Objective: To estimate the association between source-specific ambient particulate matter concentrations and lung cancer incidence in a northern Sweden cohort. Methods: Participants in the Västerbotten Intervention Programme cohort from Northern Sweden were recruited between January 1990 and December 2014 and followed until lung cancer diagnosis. Exposure to total particulate matter with aerodynamic diameter ≤ 10 µm (PM₁₀) and ≤ 2.5 µm (PM_2.5) as well as source-specific particulate matter (PM) concentrations from traffic (PM₁₀-traffic), exhaust (PM_2.5-exhaust) and wood burning (PM_2.5-wood burning) was estimated at each participant’s home address using dispersion models with high spatial resolution (down to 35 × 35 m²). Cox regression models were used to assess hazard ratios (HRs) and 95% Confidence Intervals (CIs) for the association between ambient particulate matter exposure and lung cancer incidence, adjusted for relevant potential confounding factors. Results: During 421,466 person-years of follow-up, 253 incident cases of lung cancer were observed. The risk estimates changed considerably when adjusting for individual-level baseline covariates and area level socioeconomic status. In adjusted two-pollutant models, HRs for incident lung cancer in the highest exposure tertile (tertile 3) compared to the lowest (tertile 1) were 1.15 (95% CI: 0.77–1.71) for PM₁₀-traffic, 1.15 (95% CI: 0.76–1.74) for PM_2.5-exhaust, and 1.15 (95% CI: 0.77–1.71) for PM_2.5-wood burning. Conclusions: Increased risks for lung cancer were estimated to be associated with long-term exposure to PM₁₀-traffic, PM_2.5-exhaust, and PM_2.5-wood burning. However, since these associations were not statistically significant and did not always appear linear, further studies are needed.