OBJECTIVE: To investigate relationships between mortality and circulating 25-hydroxyvitamin D (25(OH)D), 25-hydroxycholecalciferol (25(OH)D3) and 25-hydroxyergocalciferol (25(OH)D2). DESIGN: Case-cohort study within the Melbourne Collaborative Cohort Study (MCCS). We measured 25(OH)D2 and 25(OH)D3 in archived dried blood spots by LC-MS/MS. Cox regression was used to estimate mortality hazard ratios (HR), with adjustment for confounders. SETTING: General community. SUBJECTS: The MCCS included 29 206 participants, who at recruitment in 1990-1994 were aged 40-69 years, had dried blood spots collected and no history of cancer. For the present study we selected participants who died by 31 December 2007 (n 2410) and a random sample (sub-cohort, n 2996). RESULTS: The HR per 25 nmol/l increment in concentration of 25(OH)D and 25(OH)D3 were 0·86 (95 % CI 0·78, 0·96; P=0·007) and 0·85 (95 % CI 0·77, 0·95; P=0·003), respectively. Of 5108 participants, sixty-three (1·2 %) had detectable 25(OH)D2; their mean 25(OH)D concentration was 11·9 (95 % CI 7·3, 16·6) nmol/l higher (P<0·001). The HR for detectable 25(OH)D2 was 1·80 (95 % CI 1·09, 2·97; P=0·023); for those with detectable 25(OH)D2, the HR per 25 nmol/l increment in 25(OH)D was 1·06 (95 % CI 0·87, 1·29; P interaction=0·02). HR were similar for participants who reported being in good, very good or excellent health four years after recruitment. CONCLUSIONS: Total 25(OH)D and 25(OH)D3 concentrations were inversely associated with mortality. The finding that the inverse association for 25(OH)D was restricted to those with no detectable 25(OH)D2 requires confirmation in populations with higher exposure to ergocalciferol.