Four different extraction procedure representative of methods commonly employed in the isolation of somatostatin like immunoreactivity (SLI) were tested for their ability to extract large MW forms of SLI from porcine, canine and human pancreas. The yield of SLI and recovery of added somatostatin was much higher with methods involving traditional acid/ethanol extraction (methods I and II) than with methods involving boiling of tissues in water or 2 m CH3COOH (methods III and IV). Porcine and canine pancreases extracted by methods III and IV (but not methods I and II) revealed remarkable molecular heterogeneity upon gel filtration, but immuno-affinity-chromatography eliminated the largest forms. A component of approximately 3000 daltons was immunoabsorbable and resisted refiltration in 8 m urea. No large forms were detectable in human pancreases. The SLI peaks eluting at the position of synthetic somatostatin could be resolved into two components, one of which was lacking C-terminal immunoreactivity. It is concluded that the method of extraction as well as the species investigated and the specificity of the antisera employed will influence significantly the results of studies of the tissue forms of somatostatin.