Whether impaired kidney function is associated with increased risk of developing dementia is unclear. We investigated the association between estimated glomerular filtration rate (eGFR) and dementia. Using a triangulation approach, we performed (1) a prospective study in 90,369 Danes from the Copenhagen General Population Study (CGPS), (2) a meta-analysis in 468,699 Scandinavians (including CGPS) and (3) a two-sample Mendelian randomization study in 218,792-1,004,040 Europeans using summary data from largest publicly available genome wide association studies (GWASs). During up to 15 years of follow-up (CGPS), 2,468 individuals developed dementia. Age and sex standardized percentile of eGFR below versus above the median conferred a multifactorially adjusted hazard ratio of 1.09 (95% confidence interval: 1.01–1.18). In meta-analysis, random-effects risk of dementia was 1.14 (1.06–1.22) for mildly decreased eGFR (60–90 mL/min/1.73 m²), 1.31 (0.92–1.87) for moderately decreased eGFR (30–59 mL/min/1.73 m²) and 1.91 (1.21–3.01) for severely decreased eGFR (< 30 mL/min/1.73 m²), compared to reference eGFR (> 90 mL/min/1.73 m²). Using directly comparable eGFR measures (log[eGFR] scaled to one standard deviation, as well as eGFR below versus above 60 mL/min/1.73 m²), we found no association with risk of dementia in observational CGPS or in Mendelian randomization analyses. In conclusion, impaired kidney function was associated with modestly increased risk of developing dementia. This was not supported by causal, genetic analyses using a Mendelian randomization approach. However, future stronger genetic instruments for kidney function and larger GWASs with more dementia cases, particularly for the vascular dementia subtype, warrant a re-evaluation of the causal hypothesis.