The exchanges between pools of particulate DMSP (pDMSP) and the total dissolved pool of DMSP and DMS [dDMS(P)], as well as dDMS(P) removal rates were investigated in 4 shipboard incubations of amended and size-fractionated natural planktonic assemblages of the Mediterranean Sea in spring 1995 and in a laboratory experiment with cultured populations. In the shipboard experiments, the effects of different concentrations of copepods and the presence/absence of micrograzers were assessed. Removal rates of dDMS(P), obtained from seawater samples spiked with dDMSP (dissolved DMSP), were linearly correlated with dDMS(P) levels in the range 10 to 50 nM and were unrelated to size-fractionation treatments. The biological turnover rate constant of dDMS(P) was 0.5 d(-1). In most of the experiments, production of dDMS(P) was independent of copepod concentration and lowest in waters from which both copepods and micrograzers > 10 mu m had been removed. Overall, the results of the shipboard experiments suggested that (1) dDMS(P) production occurred in the microplanktonic food web, probably because pDMSP occurred predominately in the size fraction < 10 mu m and was unavailable for direct copepod consumption, and (2) dDMS(P) removal rates were likely due to organisms < 10 pm in size, probably bacteria. The laboratory experiment involved a common Mediterranean microplanktonic ciliate species (Strombidium sulcatum) grazing on a DMSP-containing Prymnesiophyceae (Isochrysis galbana). The rate of dDMS(P) release from algal DMSP was greatly increased when the phytoplankton was subjected to grazing by ciliates. The majority of prey pDMSP (> 66%) was released to solution.