BACKGROUND AND AIMS: Varenicline and nicotine replacement therapy (NRT) are the most commonly used medications to quit smoking. Given their widespread use, monitoring adverse risks remains important. This study aimed to estimate the neuropsychiatric and cardiovascular risks associated with varenicline and NRT as used in routine UK care. DESIGN: Case-cross-over study. SETTING: UK-based electronic primary care records in the Clinical Practice Research Datalink from 2006 to 2015 linked to hospital and mortality data sets. PARTICIPANTS: Adult smokers (n =282,429) observed during periods when exposed and not exposed to either varenicline or NRT. MEASUREMENTS: Main outcomes included suicide, self-harm, myocardial infarction (MI), all-cause death and cause-specific death [MI, chronic obstructive pulmonary disease (COPD)]. In primary analyses, conditional logistic regression was used to compare the chance of varenicline or NRT exposure during the risk period (90 days prior to the event) with the chance of exposure during an earlier single reference period (91-180 days prior to the event) or multiple 90-day reference periods to increase statistical power. FINDINGS: In the primary analyses, findings were inconclusive for the associations between varenicline and the main outcomes using a single reference period, while NRT was associated with MI [odds ratio (OR) = 1.40, 95% confidence interval (CI) = 1.18-1.67]. Using multiple reference periods, varenicline was associated with an increased risk of self-harm (OR = 1.32, 95% CI = 1.12-1.56) and suicide (OR = 3.56, 95% CI = 1.32-9.60) but a reduction in all-cause death (OR = 0.75, 95% CI = 0.61-0.93). NRT was associated with MI (OR = 1.54, 95% CI = 1.36-1.74), self-harm (OR = 1.30, 95% CI = 1.18-1.44) and deaths from MI (OR = 1.53, 95% CI = 1.11-2.10), COPD (OR = 1.33, 95% CI = 1.14-1.56) and all causes (OR = 1.28, 95% CI = 1.18-1.40) when using multiple reference periods. CONCLUSIONS: There appear to be positive associations between (1) nicotine replacement therapy (NRT) and myocardial infarction, death and risk of self-harm and (2) varenicline and increased risk of self-harm and suicide, as well as a negative association between varenicline and all-cause death. The associations may not be causal. They may reflect health changes at the time of smoking cessation (nicotine replacement therapy is prescribed for people with cardiac problems) or be associated with quit attempts (exposure to both medicines was associated with self-harm).