Background: We have previously shown in a Swedish cohort study that high levels of triglycerides (TG) and glucose and low levels of high density lipoprotein (HDL) cholesterol and apolipoprotein A-I (apoA-I) are related to increased prostate cancer (PC) risk. The current study investigates serum lipids and glucose in relation to severity of localized PC, at time of diagnosis and at metastatic progression. Methods: Between 2001-2004, 203 consecutive patients with localized PC were recruited. Levels of total cholesterol (TC), HDL and low density lipoprotein (LDL) cholesterol, TG, apoA-I, apolipoprotein B (apoB), and glucose were assessed at baseline, with a second measure for 74 patients after a median time of 10 years. Correlation coefficients and multivariate logistic and Cox proportional hazards regression was used to assess associations between metabolic components and PC severity at time of diagnosis and at metastatic progression. Results: Median age at diagnosis was 65 years (range: 40-76), with a median follow up of 8.1 years (0.4-13.9). 66 men underwent radical radiotherapy (33%), 57 radical prostatectomy (28%), 27 active surveillance (13%) and 17 watchful waiting (8%). At baseline, we found a statistically significant inverse correlation between PSA and the apoB/apoA-I, LDL/HDL, and TC/HDL ratios (r=-0.18, r=-0.17, r=-0.18 respectively; P<0.05). Gleason score was positively associated with HDL and inversely with TC/HDL (r=0.15, r=-0.14; P<0.05). There was no association between the second lipid measurements and baseline PSA, but HDL remained positively associated with baseline Gleason and TG inversely associated (r=0.24, r=-0.30; P<0.05). Age-adjusted multivariate model including TG, HDL, glucose, and hypertension showed a statistically significant association between glucose and Gleason ≥7 versus <7 (OR 1.14, 95%CI: 1.00-1.29). A similar model for PSA ≥15 versus <15 did not show any statistically significant results, whereas an equivalent Cox model for progression to metastasis also showed a weak positive association with glucose (HR 1.13, 95%CI: 0.95-1.35). Conclusions: Our study supports the hypothesis that metabolic markers such as serum lipids and glucose may be a biomarker of prostate cancer severity.