The cotton crop is the main agricultural activity in the economy of the colombian Caribbean in the second semester of the year and the main supplier of fibre to national industry for about 60 years. The objective of this work was to estimate the phenotypic, genetic and environmental correlations, between 11 agronomic characters and to make a path analysis for fibre yield. Data of agronomic evaluation of 10 genotypes of cotton in eight environments of the colombian Caribbean were used. In each environment experimental design at random complete blocks with four repetitions were used. The results indicated that genetic correlations were superior to phenotypic and environmental correlations. Fibre yield (FIY) presented the highest phenotypic, genetic and partial phenotypic correlations with ginning percentage (GP), seed-cotton yield (SCY) and boll weight (BOW) with values of r > 0,43 (P<0,01). The FIP (0,810) was the cause variable that showed the greatest direct effect on the REF. The YFI can be used as selection criteria to increase the YFI in cotton.