diff --git a/src/datasets/injectivedata.jl b/src/datasets/injectivedata.jl
index 4c97a3ab..15780a28 100644
--- a/src/datasets/injectivedata.jl
+++ b/src/datasets/injectivedata.jl
@@ -5,6 +5,7 @@ struct InjectiveData{V} <: AbstractDataset
     domain_variance::Union{Nothing,V}
     codomain_variance::Union{Nothing,V}
     name::String
+    data_mask::BitVector
 end
 
 function InjectiveData(
@@ -13,8 +14,9 @@ function InjectiveData(
     domain_variance = nothing,
     codomain_variance = nothing,
     name = "[no-name]",
+    data_mask = BitVector(fill(true, size(codomain)))
 )
-    InjectiveData(domain, codomain, domain_variance, codomain_variance, name)
+    InjectiveData(domain, codomain, domain_variance, codomain_variance, name, data_mask)
 end
 
 supports_contiguosly_binned(::Type{<:InjectiveData}) = true
@@ -27,24 +29,32 @@ function make_model_domain(::ContiguouslyBinned, dataset::InjectiveData)
     push!(domain, domain[end] + Δ)
     domain
 end
-make_objective(::ContiguouslyBinned, dataset::InjectiveData) = dataset.codomain
+make_objective(::ContiguouslyBinned, dataset::InjectiveData) = dataset.codomain[dataset.data_mask]
 
 make_model_domain(::OneToOne, dataset::InjectiveData) = dataset.domain
-make_objective(::OneToOne, dataset::InjectiveData) = dataset.codomain
+make_objective(::OneToOne, dataset::InjectiveData) = dataset.codomain[dataset.data_mask]
 
 function make_objective_variance(
     ::AbstractDataLayout,
     dataset::InjectiveData{V},
 )::V where {V}
-    if !isnothing(dataset.domain_variance)
-        dataset.codomain_variance
+    if !isnothing(dataset.codomain_variance)
+        dataset.codomain_variance[dataset.data_mask]
     else
         # todo: i dunno just something
-        1e-8 .* dataset.codomain
+        1e-8 .* dataset.codomain[dataset.data_mask]
     end
 end
 
-objective_transformer(::AbstractDataLayout, dataset::InjectiveData) = _DEFAULT_TRANSFORMER()
+function objective_transformer(::AbstractDataLayout, dataset::InjectiveData)
+    function _transformer!!(domain, objective)
+        @views objective[dataset.data_mask]
+    end
+    function _transformer!!(output, domain, objective)
+        @. output = objective[dataset.data_mask]
+    end
+    _transformer!!
+end
 
 make_label(dataset::InjectiveData) = dataset.name
 
diff --git a/test/fitting/test-fit-simple-dataset.jl b/test/fitting/test-fit-simple-dataset.jl
index 277f92ce..f8bbc0f1 100644
--- a/test/fitting/test-fit-simple-dataset.jl
+++ b/test/fitting/test-fit-simple-dataset.jl
@@ -44,3 +44,20 @@ prob = FittingProblem(model => data)
 
 result = fit(prob, LevenbergMarquadt())
 @test result.u[2] ≈ 6.1 atol = 0.1
+
+# fitting a contiguously binned dataset with some masked bins
+x = 10 .^ collect(range(-1, 2, 10))
+y = x .^ -2.0
+y_err = 0.1 .* y
+# introduce some bogus data points to ignore
+y[2:5] .= 2.0
+data = InjectiveData(x, y, codomain_variance=y_err)
+# mask out the bogus data points
+data.data_mask[2:5] .= false
+model = XS_PowerLaw(K=FitParam(1.0E-5), a=FitParam(2.0))
+prob = FittingProblem(model => data)
+@test SpectralFitting.common_support(model, data) isa SpectralFitting.ContiguouslyBinned
+result = fit(prob, LevenbergMarquadt())
+@test result.u[1] ≈ 2.55 atol = 0.01
+@test result.u[2] ≈ 3.0 atol = 0.05
+# note best fit photon index, u[2] should be 3 not 2 becuase y contains bin integrated values not the density