Lakes are important for storage of the essential micronutrient molybdenum (Mo) during its transfer from the continents to the oceans, but little is known about the major sources and sinks for Mo in lacustrine ecosystems. We studied Mo cycling in Castle Lake, a small subalpine lake in the Klamath-Siskiyou Mountains of Northern California underlain primarily by mafic and ultramafic rocks where primary productivity is limited by Mo bioavailability. The deeper water of the lake becomes dysoxic (9–90 μM dissolved oxygen) during the summer. This study was undertaken to identify the sources of Mo to Castle Lake and establish a Mo budget. We measured Mo concentrations in a suite of bulk solids (lake sediments, soils and bedrock) and aqueous samples (sediment porewaters, soil runoff, spring waters, snow and ice) from Castle Lake and its watershed. Lake sediments have elevated Mo (7–36 ppm) compared to soils and bedrock (0.2–2 ppm) and Mo/Al values were nearly two orders of magnitude higher than the crustal abundance. Sediment porewaters had higher Mo (4–15 nM) than lake water (2–4 nM), soil runoff (0.1–6.2 nM), snowmelt (⩽0.1 nM), lake ice (0.3–2.2 nM) and local spring waters (0.03–2.72 nM). Bulk lake sediments had negative δ^(98/95)Mo values, ranging from −0.5 to −1.0‰ (±0.1). We used the numerical model PROFILE to estimate the net reaction rate of Mo in the porewater. Model calculations ruled out diagenesis as a source of Mo to lake sediments; diagenetic Mo always represented ⩽5% of the total Mo content in sediment. We also ruled out dissolved Mo inputs from groundwater and watershed inflow as important sources of Mo. Two whole-lake experimental Mo additions in the 1960’s could have contributed a sizeable amount of Mo to the lake sediments, but only over a short period. Atmospheric deposition of anthropogenic Mo from extensive copper smelting that occurred south of Castle Lake from 1896 to 1919 and from major Californian urban centers today were negligible Mo sources. Mo flux from the sediments (0.4–0.5 nmol cm^(−2) yr^(−1)) was roughly equal to Mo fluxes from surface inflow and outflow, whereas Mo burial fluxes were significantly higher (11.5 nmol cm^(−2) yr^(−1)). Because dissolved Mo fluxes were minimal, and atmospheric Mo deposition was estimated to be a minor source of Mo (<1 nmol cm^(−2) yr^(−1)), the largest source of Mo is non-detrital particulate matter (∼12 nmol cm^(−2) yr^(−1)), likely a mixture of organic matter and Fe–Mn oxyhydroxides as supported by Mo isotopic data.