Our objective was to assess which environmental factors explain the variation in the distribution and abundance of fishery stocks. We used a novel approach, applying geostatistical tools to explain the spatial heterogeneity of purse seine yellowfin tuna (Thunnus albacares) catch in the western Indian Ocean. Geostatistical tools enable researchers to develop a more thorough understanding of the mechanisms controlling the spatial variation in fishery stocks. In addition, this tool is able to deal with complex correlations between spatial patterns and fishery harvest. The western Indian Ocean is one of the most productive purse seine yellowfin tuna (Thunnus albacares) fisheries in the world. We evaluated the variation in the spatial distribution of catch using geostatistical methods. In addition, we discuss the dynamic ecological processes influencing the spatial heterogeneity of catch. We used records of purse seine harvest of yellowfin tuna (Thunnus albacares) collected by the Indian Ocean Tuna Commission (IOTC). The data were summarized by month for 1°×1° areas between 1999-2004. We found that the spatial distribution of catch varied both between seasons and inter-annuals, but was largest between seasons. We obtained the semivariograms parameters and best-fitting semivariogram models from seine yellowfin tuna catch using geostatistical methods. We observed significant seasonal and inter-annual differences in the semivariogram parameters and the semivariogram models of spatial distribution of catch. The average spatial correlation distance (the ranges in geostatistical) was 1000 nautical miles and the values were smaller in winter than in summer. The best-fitting semivariogram models were primarily exponential and had a longer spatial corresponding distance and lower spatial dependence than other models. The spatial structural variance (mean value was 65.82% of total variance) was significantly higher than the random variance (mean value was 34.18% of total variance). We found that the spatial structure of catch had high spatial autocorrelation at 1°×1° areas scale. We investigated the relationship between the semivariogram parameter values and the catch of purse seine yellowfin tuna and attempted to explain the ecological dynamic processes explaining the spatial heterogeneity in catch. We found a strong, positive linear correlation between monthly catch and the sum of spatial variances (sill values), with a correlation coefficient of 0.930 (P<0.001). The monthly catch was also correlated to the south-northward (correlation coefficient=0.5055, P<0.1) and northwest-southeastward fractal dimension values, suggesting that catch was positively correlated with marine dynamic processes oriented in these two directions. Thus, catch decreased when the environmental process components intensified in these two directions. In summary, a number of external factors contribute to the spatial variation in yellowfin tuna catch in the western Indian Ocean, including marine currents, nutrients, and the thickness of the thermocline, which are influenced by the monsoonal climate and ENSO episodes. In addition, we identified internal factors such as the purse seine fishing methods and fish behavior that also affected the spatial distribution of harvest. Our results suggest that several environmental factors can be used to predict changes in the catch of purse seine yellowfin tuna in the western Indian Ocean. These included seasonal indices, which had a significant influence in the spatial distribution model; ENSO episodes; and the optimal semivariogram model. The selection of environmental factors for the yellowfin tuna stock assessment model should include consideration of vector meridian variables that influence marine environmental processes, such as ocean currents or wind fields.