Measurement of N-nitroso compounds in gastric juice by different methods has given conflicting results. In order to resolve this controversy, we have assessed endogenous nitrosation by the independent N-nitrosoproline excretion test in subjects who had previously undergone gastric juice analysis by one of these methods. Ten Polya gastrectomy, 10 pernicious anaemia and nine matched control subjects were fed 380 mg of nitrate in beetroot juice and 500 mg proline. N-nitrosoproline (N-Pro) synthesised intragastrically from these precursors, and quantitatively excreted by the kidneys, was measured in 24 hour urine samples (collection checked by creatinine clearance). N-Pro excretion (mean +/- SEM) was reduced (p less than 0.01) in pernicious anaemia (1.1 +/- 0.8 ng/day) compared with matched control (18.0 +/- 7.2 ng/day), and also tended to be lower (NS) in polya gastrectomy (3.2 +/- 2.3 ng/day). Twenty four hour intragastric pH was monitored on a separate occasion in 23 of the 29 subjects; 13 were hypoacidic (pH greater than 4 greater than 50% of 24 hours) and 10 were acidic. N-Pro yields were reduced (p less than 0.01) in the hypoacidic group (0.9 +/- 0.6 ng/day) compared with the acidic group (17.9 +/- 6.6 ng/day), and N-Pro was negatively associated with mean intragastric pH (tau = -0.53, p = 0.001). We conclude that endogenous synthesis of this specific N-nitroso compound is favoured by low rather than high pH. These results are concordant with those previously reported in gastric juice from the same subjects and suggest that nitrosation is chemically rather than bacterially mediated, contrary to the nitrosamine hypothesis of gastric carcinogenesis.