A meta-analysis of worldwide studies, found by a 10-year literature follow-up and/or by searching PubMed, was performed. Forty-two studies across 20 nations consistently demonstrated an association between schizophrenia and current smoking (weighted average odds ratio, OR=5.9; 95% confidence interval, CI 4.9--5.7). In 32 male studies across 18 nations, the weighted average OR was 7.2 (CI, 6.1--8.3). In 25 female studies across 15 nations, the weighted average OR was 3.3 (CI, 3.0--3.6). The association between schizophrenia and current smoking remained after using severe mentally ill controls (18 studies across 9 countries, weighted average OR was 1.9, CI 1.7--2.1) and controlling for other variables (3 studies, adjusted ORs ranged 2-3). Heavy smoking (6 studies across 4 countries, ORs ranged 1.9--6.4) and high nicotine dependence were more frequent in smokers with schizophrenia versus the general population. There was no consistent evidence that heavy smoking or high nicotine dependence was more frequent in smokers with schizophrenia versus severe mentally ill controls. Cessation rates were lower in schizophrenia smokers versus the general population. Schizophrenia patients had a higher prevalence of ever smoking than the general population (9 studies across 6 countries, weighted average OR=3.1, CI 2.4--3.8) and than severe mentally ill patients (5 studies across 5 countries, OR=2.0, CI 1.6--2.4). Moreover, in two studies adjusting for confounders schizophrenia patients had an increased risk of starting daily smoking than controls. Thus, people who are going to develop schizophrenia have risk factors that make them more vulnerable to start smoking.