The aim of this research was to indentify the variables which deserve more attention in experiments with passion fruit (Passiflora edulis Sims) and verify their relations with the principal dependent variable: total fruit yield. From data of four experiments with passion fruit culture it was calculated the correlation matrix between the variables and performed the diagnosis of multicollinearity, then applied the path analysis between the variables left. To identify the variables with more contribution in the total variability it was performed analysis of main components. The pulp weight and the fruit number have higher correlation with the fruits total weight. The total soluble solids rate, the average weight of fruits, length and fruit diameter, peel thickness and the pulp percentage has lower correlation with the fruit total weight of passion fruit.
Passiflora edulis; correlation coefficient; path coefficients; variability