Maize (Zea mays L.) has been the subject of several studies involving correlation coefficient estimates and path analysis. This critical review discusses some systematic errors that have been observed in estimating of correlation coefficients and its possible impacts on accuracy of path analysis. In a first moment, an approach about the maize crop, origin, characteristics and biometric models commonly used in genetic breeding of this crop is presented. Some obstacles found in estimates of path coefficients and the methods used to adjust them are discussed. We also present evidences and a theoretical explanation that some data arrangement methods currently used, may be overestimating the correlation coefficients in scientific studies. Data from a literature search revealing the accuracy of path analysis of some research are presented and discussed. In a last moment, we present a future perspective about how the correct estimate of the correlation coefficients may improve the accuracy of path analysis, underscoring the need for research directed to this objective.