Hamster liver microsomal membranes catalyse the synthesis of retinyl phosphate mannose (Ret-P-Man) from GDP-mannose and exogenous retinyl phosphate (Ret-P). We have previously shown that maximal Ret-P-Man synthesis occurs in vitro at 20-30 min, followed by a subsequent loss of mannose from Ret-P-Man, suggestive of an intermediary function of Ret-P-Man and/or Ret-P-Man breakdown [Shidoji, Silverman-Jones & De Luca (1982) Biochem. J. 208, 865-868; Creek, Morre, Silverman-Jones, Shidoji & De Luca (1983) Biochem. J. 210, 541-547). To monitor Ret-P-Man synthesis and breakdown carefully, we developed a chromatographic system in which mannose, Ret-P-Man, mannose phosphate and GDP-mannose are separated in a single analysis on a Mono Q column eluted with a gradient of NaCl. Using this chromatographic system, we have determined that 80-90% of the Ret-P-Man made in vitro by hamster liver membranes in 30 min is recovered with the membranes upon centrifugation. Subsequent incubation of Ret-P-Man-loaded membranes at 37 degrees C results in a non-enzymic breakdown of Ret-P-Man to beta-mannopyranosyl phosphate and anhydroretinol. However, incubation of the Ret-P-Man-loaded hamster liver membranes with GDP, but not GMP, ADP, CDP or UDP, results in a loss of mannose from Ret-P-Man and the formation of GDP-mannose and Ret-P. These results demonstrate that Ret-P-Man synthesized in vitro is subject to non-enzymic breakdown to beta-mannopyranosyl phosphate and anhydroretinol and that the GDP-mannose:retinyl phosphate mannosyltransferase reaction is reversible.