Aims Lifestyle interventions are an important and viable approach for preventing cognitive deficits. However, the results of studies on alcohol, coffee and tea consumption in relation to cognitive decline have been divergent, likely due to confounds from dose–response effects. This meta-analysis aimed to find the dose–response relationship between alcohol, coffee or tea consumption and cognitive deficits. Methods Prospective cohort studies or nested case-control studies in a cohort investigating the risk factors of cognitive deficits were searched in PubMed, Embase, the Cochrane and Web of Science up to 4th June 2020. Two authors searched the databases and extracted the data independently. We also assessed the quality of the studies with the Newcastle-Ottawa scale. Stata 15.0 software was used to perform model estimation and plot the linear or nonlinear dose–response relationship graphs. Results The search identified 29 prospective studies from America, Japan, China and some European countries. The dose–response relationships showed that compared to non-drinkers, low consumption (<11 g/day) of alcohol could reduce the risk of cognitive deficits or only dementias, but there was no significant effect of heavier drinking (>11 g/day). Low consumption of coffee reduced the risk of any cognitive deficit (<2.8 cups/day) or dementia (<2.3 cups/day). Green tea consumption was a significant protective factor for cognitive health (relative risk, 0.94; 95% confidence intervals, 0.92–0.97), with one cup of tea per day brings a 6% reduction in risk of cognitive deficits. Conclusions Light consumption of alcohol (<11 g/day) and coffee (<2.8 cups/day) was associated with reduced risk of cognitive deficits. Cognitive benefits of green tea consumption increased with the daily consumption.