Purpose: Fluid intelligence (FI), a core cognitive function involved in reasoning and problem-solving, has been linked to psychiatric outcomes in observational studies. However, causal inferences remain challenging due to confounding and reverse causation. This study aimed to investigate the causal effects of FI on 10 major psychiatric disorders using Mendelian randomization (MR). Methods: We performed a phenome-wide association study-based two-sample MR analysis to evaluate the effects of genetically proxied FI on 10 psychiatric disorders. Genome-wide association study summary statistics were obtained from large-scale European cohort studies. The inverse-variance-weighted method served as the primary analytical method, and complementary sensitivity analyses were conducted to assess pleiotropy and heterogeneity. Results: Higher FI was causally associated with a reduced risk of attention-deficit/hyperactivity disorder (ADHD) (odds ratio [OR], 0.72; P< 0.001), schizophrenia (OR, 0.69; P= 0.016), and post-traumatic stress disorder (PTSD) (OR, 0.93; P= 0.006). In contrast, FI was positively associated with an increased risk of autism spectrum disorder (ASD; OR, 1.43; P < 0.001) and anorexia nervosa (OR, 1.39; P= 0.005). No significant associations were observed for major depressive disorder, anxiety disorder, bipolar disorder, obsessive-compulsive disorder, or postpartum depression. Sensitivity analyses supported the robustness of these findings, with no evidence of directional pleiotropy or undue influence of individual single-nucleotide polymorphisms. Conclusion This study provides genetic evidence that higher FI exerts divergent causal effects across psychiatric disorders, protective in externalizing and stress-related conditions (ADHD, PTSD, and schizophrenia) but associated with an increased risk of neurodevelopmental and internalizing conditions (ASD and anorexia nervosa). These findings highlight the need for a nuanced understanding of cognitive capacity in psychiatric vulnerability and guide preventive strategies tailored to disorder-specific cognitive profiles.