Environmental context Microscopic marine organisms have the potential to influence the global climate through the production of a trace gas, dimethylsulfide, which contributes to cloud formation. Using 3 years of observations, we investigated the environmental drivers behind the production and degradation of dimethylsulfide and its precursor dimethylsulfoniopropionate. Our results highlight the important role of the microbial community in rapidly cycling these compounds and provide an important dataset for future modelling studies. Abstract Oceanic biogeochemical cycling of dimethylsulfide (DMS), and its precursor dimethylsulfoniopropionate (DMSP), has gained considerable attention over the past three decades because of the potential role of DMS in climate mediation. Here we report 3 years of monthly vertical profiles of organic sulfur cycle concentrations (DMS, particulate DMSP (DMSPp) and dissolved DMSP (DMSPd)) and rates (DMSPd consumption, biological DMS consumption and DMS photolysis) from the Bermuda Atlantic Time-series Study (BATS) site taken between 2005 and 2008. Concentrations confirm the summer paradox with mixed layer DMS peaking ~90 days after peak DMSPp and ~50 days after peak DMSPp:Chl. A small decline in mixed layer DMS was observed relative to those measured during a previous study at BATS (1992–1994), potentially driven by long-term climate shifts at the site. On average, DMS cycling occurred on longer timescales than DMSPd (0.43±0.35 v. 1.39±0.76 day–1) with DMSPd consumption rates remaining elevated throughout the year despite significant seasonal variability in the bacterial DMSP degrader community. DMSPp was estimated to account for 4–5% of mixed layer primary production and turned over at a significantly slower rate (~0.2 day–1). Photolysis drove DMS loss in the mixed layer during the summer, whereas biological consumption of DMS was the dominant loss process in the winter and at depth. These findings offer new insight into the underlying mechanisms driving DMS(P) cycling in the oligotrophic ocean, provide an extended dataset for future model evaluation and hypothesis testing and highlight the need for a reexamination of past modelling results and conclusions drawn from data collected with old methodologies.