Abstract Background Omega-3 polyunsaturated fatty acids (omega-3 PUFAs) exhibit potential as therapeutics for a variety of diseases. This observational and Mendelian randomization (MR) study aims to explore the relationship between omega-3 PUFAs and osteoarthritis (OA). Methods Excluding individuals under 20 years old and those with missing data on relevant variables in the National Health and Nutrition Examination Survey (NHANES) spanning from 2003 to 2016, a total of 22 834 participants were included in this cross-sectional study. Weighted multivariable-adjusted logistic regression was used to estimate the association between omega-3 PUFAs and OA in adults. Moreover, restricted cubic splines were utilized to examine the dose-response relationship between omega-3 PUFAs and OA. To further investigate the potential causal relationship between omega-3 PUFAs and OA risk, a two-sample MR study was conducted. Furthermore, the robustness of the findings was assessed using various methods. Results Omega-3 PUFAs intake were inversely associated with OA in adults aged 40 ∼ 59 after multivariable adjustment $$[\text{OR} (95\% \text{CI): }0.85 (0.73, 0.98), P = 0.027]$$ , with a nonlinear relationship observed between omega-3 PUFAs intake and OA $$\left(P \text{ for non-linearity}\text{ = 0.034}\right)$$ . The IVW results showed there was no evidence to suggest a causal relationship between omega-3 PUFAs and OA risk $$\text{[OR} (95\% \text{CI): }0.967 (0.863, 1.084), P = 0.568]$$ . Conclusions Omega-3 PUFAs were inversely associated with OA in adults aged 40 ∼ 59. However, MR studies did not confirm a causal relationship between the two.