Distinguishing between natural and anthropic oil slicks is considered a challenging task, especially in Gulf of Mexico where these events can be simultaneously observed and recognized as seeps or spills. In this context, the powerful data analysis provided by Machine Learning (ML) methods was employed to develop, test, and implement a classification model (CM) to distinguish the oil slick source (OSS). From a database containing 4,916 oil samples, detected using Synthetic Aperture Radar (SAR), there were 408 spectral and 10 geometric features. The exploratory data analysis successfully reduced the dataset without compromising the model accuracy, selecting 12 features for the CM designing. An innovative approach evaluated how external factors like seasonality limit or improve the OSS predictions. To accomplish this, specific classification models (SCM) were derived from the global, tuning the best algorithms and parameters according to different scenarios. The median accuracies results revealed winter and spring as the best seasons. Among the six tested ML algorithms, Random Forest (RF) was the most robust, performing better in more than half of the investigated scenarios. The global CM achieved 73.15% of maximum accuracy using RF. The accuracy increment provided by the well-fitted models may minimize the confusion between seeps and spills. This represents a concrete contribution to reduce economic and geologic risks derived from exploration activities in offshore areas. Additionally, from the operational standpoint, specific models support specialists to select the best seasons for new acquisitions, as well as to optimize performances according to available data.