diff --git a/src/FixedPointNumbers.jl b/src/FixedPointNumbers.jl
index 970cd8cb..7d0092bc 100644
--- a/src/FixedPointNumbers.jl
+++ b/src/FixedPointNumbers.jl
@@ -200,9 +200,19 @@ saturating_sub(x::X, y::X) where {X <: FixedPoint} =
 saturating_sub(x::X, y::X) where {X <: FixedPoint{<:Unsigned}} = X(x.i - min(x.i, y.i), 0)
 
 # checked arithmetic
-checked_neg(x::X) where {X <: FixedPoint} = X(checked_neg(x.i), 0)
-checked_add(x::X, y::X) where {X <: FixedPoint} = X(checked_add(x.i, y.i), 0)
-checked_sub(x::X, y::X) where {X <: FixedPoint} = X(checked_sub(x.i, y.i), 0)
+checked_neg(x::X) where {X <: FixedPoint} = checked_sub(zero(X), x)
+function checked_add(x::X, y::X) where {X <: FixedPoint}
+    r, f = Base.Checked.add_with_overflow(x.i, y.i)
+    z = X(r, 0) # store first
+    f && throw_overflowerror(:+, x, y)
+    z
+end
+function checked_sub(x::X, y::X) where {X <: FixedPoint}
+    r, f = Base.Checked.sub_with_overflow(x.i, y.i)
+    z = X(r, 0) # store first
+    f && throw_overflowerror(:-, x, y)
+    z
+end
 
 # default arithmetic
 const DEFAULT_ARITHMETIC = :wrapping
@@ -424,6 +434,13 @@ scaledual(::Type{Tdual}, x::AbstractArray{T}) where {Tdual, T <: FixedPoint} =
     throw(ArgumentError(String(take!(io))))
 end
 
+@noinline function throw_overflowerror(op::Symbol, @nospecialize(x), @nospecialize(y))
+    io = IOBuffer()
+    print(io, x, ' ', op, ' ', y, " overflowed for type ")
+    showtype(io, typeof(x))
+    throw(OverflowError(String(take!(io))))
+end
+
 function Random.rand(r::AbstractRNG, ::SamplerType{X}) where X <: FixedPoint
     X(rand(r, rawtype(X)), 0)
 end
diff --git a/test/fixed.jl b/test/fixed.jl
index 251934b8..205fb9a0 100644
--- a/test/fixed.jl
+++ b/test/fixed.jl
@@ -275,8 +275,8 @@ end
         xs = typemin(F):eps(F):typemax(F)
         fneg(x) = -float(x)
         @test all(x -> wrapping_neg(wrapping_neg(x)) === x, xs)
-        @test all(x -> saturating_neg(x) == clamp(fneg(x), F), xs)
-        @test all(x -> !(typemin(F) < fneg(x) < typemax(F)) ||
+        @test all(x -> saturating_neg(x) === clamp(fneg(x), F), xs)
+        @test all(x -> !(typemin(F) <= fneg(x) <= typemax(F)) ||
                        wrapping_neg(x) === checked_neg(x) === fneg(x) % F, xs)
     end
 end
@@ -301,8 +301,8 @@ end
         xys = ((x, y) for x in xs, y in xs)
         fadd(x, y) = float(x) + float(y)
         @test all(((x, y),) -> wrapping_sub(wrapping_add(x, y), y) === x, xys)
-        @test all(((x, y),) -> saturating_add(x, y) == clamp(fadd(x, y), F), xys)
-        @test all(((x, y),) -> !(typemin(F) < fadd(x, y) < typemax(F)) ||
+        @test all(((x, y),) -> saturating_add(x, y) === clamp(fadd(x, y), F), xys)
+        @test all(((x, y),) -> !(typemin(F) <= fadd(x, y) <= typemax(F)) ||
                                wrapping_add(x, y) === checked_add(x, y) === fadd(x, y) % F, xys)
     end
 end
diff --git a/test/normed.jl b/test/normed.jl
index 6c751724..c2e1c6e3 100644
--- a/test/normed.jl
+++ b/test/normed.jl
@@ -303,7 +303,7 @@ end
         fneg(x) = -float(x)
         @test all(x -> wrapping_neg(wrapping_neg(x)) === x, xs)
         @test all(x -> saturating_neg(x) === clamp(fneg(x), N), xs)
-        @test all(x -> !(typemin(N) < fneg(x) < typemax(N)) ||
+        @test all(x -> !(typemin(N) <= fneg(x) <= typemax(N)) ||
                        wrapping_neg(x) === checked_neg(x) === fneg(x) % N, xs)
     end
 end
@@ -329,7 +329,7 @@ end
         fadd(x, y) = float(x) + float(y)
         @test all(((x, y),) -> wrapping_sub(wrapping_add(x, y), y) === x, xys)
         @test all(((x, y),) -> saturating_add(x, y) === clamp(fadd(x, y), N), xys)
-        @test all(((x, y),) -> !(typemin(N) < fadd(x, y) < typemax(N)) ||
+        @test all(((x, y),) -> !(typemin(N) <= fadd(x, y) <= typemax(N)) ||
                                wrapping_add(x, y) === checked_add(x, y) === fadd(x, y) % N, xys)
     end
 end
@@ -354,7 +354,7 @@ end
         fsub(x, y) = float(x) - float(y)
         @test all(((x, y),) -> wrapping_add(wrapping_sub(x, y), y) === x, xys)
         @test all(((x, y),) -> saturating_sub(x, y) === clamp(fsub(x, y), N), xys)
-        @test all(((x, y),) -> !(typemin(N) < fsub(x, y) < typemax(N)) ||
+        @test all(((x, y),) -> !(typemin(N) <= fsub(x, y) <= typemax(N)) ||
                                wrapping_sub(x, y) === checked_sub(x, y) === fsub(x, y) % N, xys)
     end
 end