With the possible most powerful goodness-of-fit test available until now, the Halperin's chi-square, a statistically significant association between schizophrenia and birth order in 1016 brazilian patients is reported. This association is with early and last born males, favoring theories which put emphasis on stressful experiences over children who are in the extremes of their sibships. There is a sex factor: females show no association. Factors which could introduce bias, as family size, social class and incomplete sibships were tentatively controlled. Since in Brazil no population data are available, the Greenwood-Yule method seems useful and valid for the purpose, with the aid provided by Halperin's chi square modification. A brief comment on results and earlier studies is made, together with suggestions for future research.