PR-104A is a clinical-stage nitrogen mustard prodrug that is activated for DNA alkylation by reduction of a nitro group to the corresponding hydroxylamine (PR-104H) or amine (PR-104M). Metabolic reduction is catalysed by flavoreductases such as cytochrome P450 oxidoreductase (POR) under hypoxia, or by aldo-ketoreductase 1C3 (AKR1C3) independently of hypoxia. The unstable reduced metabolites are challenging to measure in biological samples, and biomarkers of the metabolic activation of PR-104A have not been used in the clinical evaluation of PR-104 to date. Here, we employ a selected reaction monitoring mass spectrometry assay for DNA crosslinks to assess the capacity of human cancer cells to bioactivate PR-104A. We also test whether the more abundant DNA monoadducts could be used for the same purpose. DNA monoadducts and crosslinks from PR-104A itself, and from its reduced metabolites, accumulated over 4 h in AKR1C3-expressing TF1 erythroleukaemia cells under hypoxia, whereas intracellular concentrations of unstable PR-104H and PR-104M reached steady state within 1 h. We then varied rates of PR-104A reduction by manipulating hypoxia or reductase expression in a panel of cell lines, in which AKR1C3 and POR were quantified by targeted proteomics. Hypoxia or reductase overexpression induced large increases in PR-104A sensitivity (inhibition of proliferation), DNA damage response (γH2AX formation), steady-state concentrations of PR-104H/M and formation of reduced drug-DNA adducts but not DNA adducts retaining the dinitro groups of PR-104A. The fold-change in the sum of PR-104H and PR-104M correlated with the fold-change in reduced crosslinks or monoadducts (R2 = 0.87 for both), demonstrating their potential for assessing the capacity of cancer cells to bioactivate PR-104A.