Two gastrin antisera with different specificity for gastrins I and II have been compared by radioimmunoassay. Although both were produced in rabbits in response to repeated injections of synthetic human gastrin I (SHG I), one (AS 4) binds well with SHG I and porcine gastrin I (PG I), yet shows little cross-reactivity with porcine gastrin II (PG II), whilst the other (AS 35) shows equal binding with porcine gastrins I and II and SHG I. Fasting serum gastrin levels in normals, patients with duodenal ulcer, gastric ulcer, pernicious anaemia and Zollinger-Ellison syndrome are higher with AS 35 than with AS 4. In particular, levels in duodenal ulcer, which are significantly lower than normal with AS 4, are normal or slightly higher than normal with AS 35. In general, responses of the various groups to protein feeding, insulin hypoglycaemia and secretin injection have been in the same direction with both antisera. The exceptions are some patients who had a complete truncal vagotomy whose serum gastrin in response to insulin hypoglycaemia was unchanged with AS 4, but showed a small rise with AS 35. These studies perhaps offer one explanation for the variations in serum gastrin levels obtained in different laboratories and may theoretically suggest that the form in which gastrin is secreted in duodenal ulcer disease may be as sulphated gastrin II rather than as non-sulphated gastrin I. Antibodies raised against gastrin II may shed more light on this hypothesis.