Associations of two dichotomous variables can be expressed by a variety of measures. Relative risk, odds ratio and risk reductions belong to the commonly used ones. By illustrations it is shown that the odds ratio differs markedly from the relative risk when two conditions are fulfilled: One rate is close to 50% while the other is not. The odds ratio generally is higher than the relative risk, hence taking it as an approximation for the relative risk may overestimate effects under certain circumstances. This may particularly affect meta analysis of binary data, where the odds ratio is a frequently used measure. On the other hand, relative risk, relative and absolute risk reduction are dependent on the way the four fold table was arranged � while the odds ratio is not. A formula for transposing the odds ratio into the relative risk is provided.

relative risk, odds ratio, relative risk reduction, absolute risk reduction.