diff --git a/NEWS.md b/NEWS.md
index 9bf97ddb3b..d83e76d62a 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,5 +1,8 @@
 # Flux Release Notes
 
+## v0.13.7
+* Added [`@autosize` macro](https://github.com/FluxML/Flux.jl/pull/2078)
+
 ## v0.13.4
 * Added [`PairwiseFusion` layer](https://github.com/FluxML/Flux.jl/pull/1983)
 
diff --git a/docs/src/outputsize.md b/docs/src/outputsize.md
index d692816b46..9376db9ab8 100644
--- a/docs/src/outputsize.md
+++ b/docs/src/outputsize.md
@@ -1,47 +1,79 @@
 # Shape Inference
 
-To help you generate models in an automated fashion, [`Flux.outputsize`](@ref) lets you 
-calculate the size returned produced by layers for a given size input.
-This is especially useful for layers like [`Conv`](@ref).
+Flux has some tools to help generate models in an automated fashion, by inferring the size
+of arrays that layers will recieve, without doing any computation. 
+This is especially useful for convolutional models, where the same [`Conv`](@ref) layer
+accepts any size of image, but the next layer may not. 
 
-It works by passing a "dummy" array into the model that preserves size information without running any computation.
-`outputsize(f, inputsize)` works for all layers (including custom layers) out of the box.
-By default, `inputsize` expects the batch dimension,
-but you can exclude the batch size with `outputsize(f, inputsize; padbatch=true)` (assuming it to be one).
+The higher-level tool is a macro [`@autosize`](@ref) which acts on the code defining the layers,
+and replaces each appearance of `_` with the relevant size. This simple example returns a model
+with `Dense(845 => 10)` as the last layer:
 
-Using this utility function lets you automate model building for various inputs like so:
 ```julia
-"""
-    make_model(width, height, inchannels, nclasses;
-               layer_config = [16, 16, 32, 32, 64, 64])
+@autosize (28, 28, 1, 32) Chain(Conv((3, 3), _ => 5, relu, stride=2), Flux.flatten, Dense(_ => 10))
+```
+
+The input size may be provided at runtime, like `@autosize (sz..., 1, 32) Chain(Conv(`..., but all the
+layer constructors containing `_` must be explicitly written out -- the macro sees the code as written.
+
+This macro relies on a lower-level function [`outputsize`](@ref Flux.outputsize), which you can also use directly:
+
+```julia
+c = Conv((3, 3), 1 => 5, relu, stride=2)
+Flux.outputsize(c, (28, 28, 1, 32))  # returns (13, 13, 5, 32)
+```
 
-Create a CNN for a given set of configuration parameters.
+The function `outputsize` works by passing a "dummy" array into the model, which propagates through very cheaply.
+It should work for all layers, including custom layers, out of the box.
 
-# Arguments
-- `width`: the input image width
-- `height`: the input image height
-- `inchannels`: the number of channels in the input image
-- `nclasses`: the number of output classes
-- `layer_config`: a vector of the number of filters per each conv layer
+An example of how to automate model building is this:
+```jldoctest; output = false, setup = :(using Flux)
 """
-function make_model(width, height, inchannels, nclasses;
-                    layer_config = [16, 16, 32, 32, 64, 64])
-  # construct a vector of conv layers programmatically
-  conv_layers = [Conv((3, 3), inchannels => layer_config[1])]
-  for (infilters, outfilters) in zip(layer_config, layer_config[2:end])
-    push!(conv_layers, Conv((3, 3), infilters => outfilters))
+    make_model(width, height, [inchannels, nclasses; layer_config])
+
+Create a CNN for a given set of configuration parameters. Arguments:
+- `width`, `height`: the input image size in pixels
+- `inchannels`: the number of channels in the input image, default `1`
+- `nclasses`: the number of output classes, default `10`
+- Keyword `layer_config`: a vector of the number of channels per layer, default `[16, 16, 32, 64]`
+"""
+function make_model(width, height, inchannels = 1, nclasses = 10;
+                    layer_config = [16, 16, 32, 64])
+  # construct a vector of layers:
+  conv_layers = []
+  push!(conv_layers, Conv((5, 5), inchannels => layer_config[1], relu, pad=SamePad()))
+  for (inch, outch) in zip(layer_config, layer_config[2:end])
+    push!(conv_layers, Conv((3, 3), inch => outch, sigmoid, stride=2))
   end
 
-  # compute the output dimensions for the conv layers
-  # use padbatch=true to set the batch dimension to 1
-  conv_outsize = Flux.outputsize(conv_layers, (width, height, nchannels); padbatch=true)
+  # compute the output dimensions after these conv layers:
+  conv_outsize = Flux.outputsize(conv_layers, (width, height, inchannels); padbatch=true)
 
-  # the input dimension to Dense is programatically calculated from
-  #  width, height, and nchannels
-  return Chain(conv_layers..., Dense(prod(conv_outsize) => nclasses))
+  # use this to define appropriate Dense layer:
+  last_layer = Dense(prod(conv_outsize) => nclasses)
+  return Chain(conv_layers..., Flux.flatten, last_layer)
 end
+
+m = make_model(28, 28, 3, layer_config = [9, 17, 33, 65])
+
+Flux.outputsize(m, (28, 28, 3, 42)) == (10, 42) == size(m(randn(Float32, 28, 28, 3, 42)))
+
+# output
+
+true
 ```
 
+Alternatively, using the macro, the definition of `make_model` could end with:
+
+```
+  # compute the output dimensions & construct appropriate Dense layer:
+  return @autosize (width, height, inchannels, 1) Chain(conv_layers..., Flux.flatten, Dense(_ => nclasses))
+end
+```
+
+### Listing
+
 ```@docs
+Flux.@autosize
 Flux.outputsize
 ```
diff --git a/src/Flux.jl b/src/Flux.jl
index 5578e94f08..fcb473ba2c 100644
--- a/src/Flux.jl
+++ b/src/Flux.jl
@@ -55,6 +55,7 @@ include("layers/show.jl")
 include("loading.jl")
 
 include("outputsize.jl")
+export @autosize
 
 include("data/Data.jl")
 using .Data
diff --git a/src/outputsize.jl b/src/outputsize.jl
index 774b75ff26..40170ad8d2 100644
--- a/src/outputsize.jl
+++ b/src/outputsize.jl
@@ -147,8 +147,12 @@ outputsize(m::AbstractVector, input::Tuple...; padbatch=false) = outputsize(Chai
 
 ## bypass statistics in normalization layers
 
-for layer in (:LayerNorm, :BatchNorm, :InstanceNorm, :GroupNorm)
-  @eval (l::$layer)(x::AbstractArray{Nil}) = x
+for layer in (:BatchNorm, :InstanceNorm, :GroupNorm)  # LayerNorm works fine
+  @eval function (l::$layer)(x::AbstractArray{Nil})
+    l.chs == size(x, ndims(x)-1) || throw(DimensionMismatch(
+      string($layer, " expected ", l.chs, " channels, but got size(x) == ", size(x))))
+    x
+  end
 end
 
 ## fixes for layers that don't work out of the box
@@ -168,3 +172,162 @@ for (fn, Dims) in ((:conv, DenseConvDims),)
     end
   end
 end
+
+
+"""
+    @autosize (size...,) Chain(Layer(_ => 2), Layer(_), ...)
+
+Returns the specified model, with each `_` replaced by an inferred number,
+for input of the given `size`.
+
+The unknown sizes are usually the second-last dimension of that layer's input,
+which Flux regards as the channel dimension.
+(A few layers, `Dense` & [`LayerNorm`](@ref), instead always use the first dimension.)
+The underscore may appear as an argument of a layer, or inside a `=>`.
+It may be used in further calculations, such as `Dense(_ => _÷4)`.
+
+# Examples
+```
+julia> @autosize (3, 1) Chain(Dense(_ => 2, sigmoid), BatchNorm(_, affine=false))
+Chain(
+  Dense(3 => 2, σ),                     # 8 parameters
+  BatchNorm(2, affine=false),
+) 
+
+julia> img = [28, 28];
+
+julia> @autosize (img..., 1, 32) Chain(              # size is only needed at runtime
+          Chain(c = Conv((3,3), _ => 5; stride=2, pad=SamePad()),
+                p = MeanPool((3,3)),
+                b = BatchNorm(_),
+                f = Flux.flatten),
+          Dense(_ => _÷4, relu, init=Flux.rand32),   # can calculate output size _÷4
+          SkipConnection(Dense(_ => _, relu), +),
+          Dense(_ => 10),
+       ) |> gpu                                      # moves to GPU after initialisation
+Chain(
+  Chain(
+    c = Conv((3, 3), 1 => 5, pad=1, stride=2),  # 50 parameters
+    p = MeanPool((3, 3)),
+    b = BatchNorm(5),                   # 10 parameters, plus 10
+    f = Flux.flatten,
+  ),
+  Dense(80 => 20, relu),                # 1_620 parameters
+  SkipConnection(
+    Dense(20 => 20, relu),              # 420 parameters
+    +,
+  ),
+  Dense(20 => 10),                      # 210 parameters
+)         # Total: 10 trainable arrays, 2_310 parameters,
+          # plus 2 non-trainable, 10 parameters, summarysize 10.469 KiB.
+
+julia> outputsize(ans, (28, 28, 1, 32))
+(10, 32)
+```
+
+Limitations:
+* While `@autosize (5, 32) Flux.Bilinear(_ => 7)` is OK, something like `Bilinear((_, _) => 7)` will fail.
+* While `Scale(_)` and `LayerNorm(_)` are fine (and use the first dimension), `Scale(_,_)` and `LayerNorm(_,_)`
+  will fail if `size(x,1) != size(x,2)`.
+* RNNs won't work: `@autosize (7, 11) LSTM(_ => 5)` fails, because `outputsize(RNN(3=>7), (3,))` also fails, a known issue.
+"""
+macro autosize(size, model)
+  Meta.isexpr(size, :tuple) || error("@autosize's first argument must be a tuple, the size of the input")
+  Meta.isexpr(model, :call) || error("@autosize's second argument must be something like Chain(layers...)")
+  ex = _makelazy(model)
+  @gensym m
+  quote
+    $m = $ex
+    $outputsize($m, $size)
+    $striplazy($m)
+  end |> esc
+end
+
+function _makelazy(ex::Expr)
+  n = _underscoredepth(ex)
+  n == 0 && return ex
+  n == 1 && error("@autosize doesn't expect an underscore here: $ex")
+  n == 2 && return :($LazyLayer($(string(ex)), $(_makefun(ex)), nothing))
+  n > 2 && return Expr(ex.head, ex.args[1], map(_makelazy, ex.args[2:end])...)
+end
+_makelazy(x) = x
+
+function _underscoredepth(ex::Expr)
+  # Meta.isexpr(ex, :tuple) && :_ in ex.args && return 10
+  ex.head in (:call, :kw, :(->), :block) || return 0
+  ex.args[1] === :(=>) && ex.args[2] === :_ && return 1
+  m = maximum(_underscoredepth, ex.args)
+  m == 0 ? 0 : m+1
+end
+_underscoredepth(ex) = Int(ex === :_)
+
+function _makefun(ex)
+  T = Meta.isexpr(ex, :call) ? ex.args[1] : Type
+  @gensym x s
+  Expr(:(->), x, Expr(:block, :($s = $autosizefor($T, $x)), _replaceunderscore(ex, s)))
+end
+
+"""
+    autosizefor(::Type, x)
+
+If an `_` in your layer's constructor, used within `@autosize`, should
+*not* mean the 2nd-last dimension, then you can overload this.
+
+For instance `autosizefor(::Type{<:Dense}, x::AbstractArray) = size(x, 1)`
+is needed to make `@autosize (2,3,4) Dense(_ => 5)` return 
+`Dense(2 => 5)` rather than `Dense(3 => 5)`.
+"""
+autosizefor(::Type, x::AbstractArray) = size(x, max(1, ndims(x)-1))
+autosizefor(::Type{<:Dense}, x::AbstractArray) = size(x, 1)
+autosizefor(::Type{<:LayerNorm}, x::AbstractArray) = size(x, 1)
+
+_replaceunderscore(e, s) = e === :_ ? s : e
+_replaceunderscore(ex::Expr, s) = Expr(ex.head, map(a -> _replaceunderscore(a, s), ex.args)...)
+
+mutable struct LazyLayer
+  str::String
+  make::Function
+  layer
+end
+
+@functor LazyLayer
+
+function (l::LazyLayer)(x::AbstractArray, ys::AbstractArray...)
+  l.layer === nothing || return l.layer(x, ys...)
+  made = l.make(x)  # for something like `Bilinear((_,__) => 7)`, perhaps need `make(xy...)`, later.
+  y = made(x, ys...)
+  l.layer = made  # mutate after we know that call worked
+  return y
+end
+
+function striplazy(m)
+  fs, re = functor(m)
+  re(map(striplazy, fs))
+end
+function striplazy(l::LazyLayer)
+  l.layer === nothing || return l.layer
+  error("LazyLayer should be initialised, e.g. by outputsize(model, size), before using stiplazy")
+end
+
+# Could make LazyLayer usable outside of @autosize, for instance allow Chain(@lazy Dense(_ => 2))?
+# But then it will survive to produce weird structural gradients etc. 
+
+function ChainRulesCore.rrule(l::LazyLayer, x)
+  l(x), _ -> error("LazyLayer should never be used within a gradient. Call striplazy(model) first to remove all.")
+end
+function ChainRulesCore.rrule(::typeof(striplazy), m)
+  striplazy(m), _ -> error("striplazy should never be used within a gradient")
+end
+
+params!(p::Params, x::LazyLayer, seen = IdSet()) = error("LazyLayer should never be used within params(m). Call striplazy(m) first.")
+function Base.show(io::IO, l::LazyLayer)
+  printstyled(io, "LazyLayer(", color=:light_black)
+  if l.layer == nothing
+    printstyled(io, l.str, color=:magenta)
+  else
+    printstyled(io, l.layer, color=:cyan)
+  end
+  printstyled(io, ")", color=:light_black)
+end
+
+_big_show(io::IO, l::LazyLayer, indent::Int=0, name=nothing) = _layer_show(io, l, indent, name)
diff --git a/test/outputsize.jl b/test/outputsize.jl
index 667b5bad76..2e9595f699 100644
--- a/test/outputsize.jl
+++ b/test/outputsize.jl
@@ -142,16 +142,103 @@ end
   m = LayerNorm(32)
   @test outputsize(m, (32, 32, 3, 16)) == (32, 32, 3, 16)
   @test outputsize(m, (32, 32, 3); padbatch=true) == (32, 32, 3, 1)
+  m2 = LayerNorm(3, 2)
+  @test outputsize(m2, (3, 2)) == (3, 2) == size(m2(randn(3, 2)))
+  @test outputsize(m2, (3,)) == (3, 2) == size(m2(randn(3, 2)))
 
   m = BatchNorm(3)
   @test outputsize(m, (32, 32, 3, 16)) == (32, 32, 3, 16)
   @test outputsize(m, (32, 32, 3); padbatch=true) == (32, 32, 3, 1)
+  @test_throws Exception m(randn(Float32, 32, 32, 5, 1))
+  @test_throws DimensionMismatch outputsize(m, (32, 32, 5, 1))
 
   m = InstanceNorm(3)
   @test outputsize(m, (32, 32, 3, 16)) == (32, 32, 3, 16)
   @test outputsize(m, (32, 32, 3); padbatch=true) == (32, 32, 3, 1)
+  @test_throws Exception m(randn(Float32, 32, 32, 5, 1))
+  @test_throws DimensionMismatch outputsize(m, (32, 32, 5, 1))
 
   m = GroupNorm(16, 4)
   @test outputsize(m, (32, 32, 16, 16)) == (32, 32, 16, 16)
   @test outputsize(m, (32, 32, 16); padbatch=true) == (32, 32, 16, 1)
+  @test_throws Exception m(randn(Float32, 32, 32, 15, 4))
+  @test_throws DimensionMismatch outputsize(m, (32, 32, 15, 4))
+end
+
+@testset "autosize macro" begin
+  m = @autosize (3,) Dense(_ => 4)
+  @test randn(3) |> m |> size == (4,)
+
+  m = @autosize (3, 1) Chain(Dense(_, 4), Dense(4 => 10), softmax)
+  @test randn(3, 5) |> m |> size == (10, 5)
+  
+  m = @autosize (2, 3, 4, 5) Dense(_ => 10)  # goes by first dim, not 2nd-last
+  @test randn(2, 3, 4, 5) |> m |> size == (10, 3, 4, 5)
+  
+  m = @autosize (9,) Dense(_ => div(_,2))
+  @test randn(9) |> m |> size == (4,)
+
+  m = @autosize (3,) Chain(one = Dense(_ => 4), two = softmax)  # needs kw
+  @test randn(3) |> m |> size == (4,)
+
+  m = @autosize (3, 45) Maxout(() -> Dense(_ => 6, tanh), 2)    # needs ->, block
+  @test randn(3, 45) |> m |> size == (6, 45)
+
+  # here Parallel gets two inputs, no problem:
+  m = @autosize (3,) Chain(SkipConnection(Dense(_ => 4), Parallel(vcat, Dense(_ => 5), Dense(_ => 6))), Flux.Scale(_))
+  @test randn(3) |> m |> size == (11,)
+  
+  # like Dense, LayerNorm goes by the first dimension:
+  m = @autosize (3, 4, 5) LayerNorm(_)
+  @test rand(3, 6, 7) |> m |> size == (3, 6, 7)
+
+  m = @autosize (3, 3, 10) LayerNorm(_, _)  # does not check that sizes match
+  @test rand(3, 3, 10) |> m |> size == (3, 3, 10)
+  
+  m = @autosize (3,) Flux.Bilinear(_ => 10)
+  @test randn(3) |> m |> size == (10,)
+
+  m = @autosize (3, 1) Flux.Bilinear(_ => 10)
+  @test randn(3, 4) |> m |> size == (10, 4)
+  
+  m = @autosize (3,) SkipConnection(Dense(_ => _), Flux.Bilinear(_ => 10))  # Bilinear gets two inputs
+  @test randn(3, 4) |> m |> size == (10, 4)
+  
+  @test_throws Exception @eval @autosize (3,) Flux.Bilinear((_,3) => 10)
+  
+  # first docstring example
+  m = @autosize (3, 1) Chain(Dense(_ => 2, sigmoid), BatchNorm(_, affine=false))
+  @test randn(3, 4) |> m |> size == (2, 4)
+  
+  # evil docstring example
+  img = [28, 28];
+  m = @autosize (img..., 1, 32) Chain(              # size is only needed at runtime
+         Chain(c = Conv((3,3), _ => 5; stride=2, pad=SamePad()),
+               p = MeanPool((3,3)),
+               b = BatchNorm(_),
+               f = Flux.flatten),
+         Dense(_ => _÷4, relu, init=Flux.rand32),   # can calculate output size _÷4
+         SkipConnection(Dense(_ => _, relu), +),
+         Dense(_ => 10),
+      ) |> gpu                                      # moves to GPU after initialisation
+  @test randn(Float32, img..., 1, 32) |> gpu |> m |> size == (10, 32)
+end
+
+@testset "LazyLayer" begin
+  # This is what `@autosize` uses, ideally nobody should make these by hand!
+  # Implicitly testeed by the macro, explicitly here too: 
+  ld = Flux.LazyLayer("Dense(_ => 3, relu; init=??)", x -> Dense(Flux.autosizefor(Dense, x) => 3, relu, init=ones), nothing)
+
+  lm = Chain(ld, Flux.Scale(3))
+  @test string(ld) == "LazyLayer(Dense(_ => 3, relu; init=??))"
+  @test_throws Exception Flux.striplazy(lm)
+
+  @test lm([1,2]) == [3,3,3]
+
+  @test string(ld) == "LazyLayer(Dense(2 => 3, relu))"
+  @test Flux.striplazy(ld) isa Dense
+
+  @test_throws Exception Flux.params(lm)
+  @test_throws Exception gradient(x -> sum(abs2, lm(x)), [1,2])
+  @test_throws Exception gradient(m -> sum(abs2, Flux.striplazy(m)([1,2])), ld)
 end