Background: The pathogenic role of hyperhomocysteinemia in cryptogenic stroke is not well established. We aimed to determine homocysteine levels in patients with cryptogenic stroke considering the possible variables that may act as confounders and analyze the influence of obesity on this association. Patients and methods: We conducted a case-control study in 123 patients with cryptogenic stroke aged 42 ± 12 years and in 153 control subjects aged 42 ± 13 years. Serum homocysteine was determined by fluorescence polarization immunoassay. Results: Patients showed statistically higher levels of homocysteine, creatinine and higher BMI than controls (p = 0.045, p = 0.014, p = 0.013), respectively. After multivariate adjustment the differences in homocysteine levels disappeared (p = 0.774). When subjects were classified according to the presence or absence of obesity, the differences in the prevalence of hyperhomocysteinemia (homocysteine >15 μM) were highly significant, being higher in patients than in controls (p = 0.009). Likewise, mean values of homocysteine in obese were higher in cases than in controls (16.9 ± 9.5 μM vs. 10.12 ± 2.5 μM; p = 0.020), remaining significant after adjusting for the above mentioned confounders. Conclusion: Although in general, hyperhomocysteinemia does not seem to constitute an independent risk factor for cryptogenic stroke, it significantly increases the risk in obese subjects; therefore it is convenient to decrease its levels in this sub-group to minimize the risk.