This article applies error propagation in a Monte Carlo simulation for a spatial-based fuzzy logic multi-criteria evaluation (MCE) in order to investigate the output uncertainty created by the input data sets and model structure. Six scenarios for quantifying uncertainty are reviewed. Three scenarios are progressively more complex in defining observational data (attribute uncertainty); while three other scenarios include uncertainty in observational data (position of boundaries between map units), weighting of evidence (fuzzy membership assignment), and evaluating changes in the MCE model (fuzzy logic operators). A case study of petroleum exploration in northern South America is used. Despite the resources and time required, the best estimate of input uncertainty is that based on expert-defined values. Uncertainties for fuzzy membership assignment and boundary transition zones do not affect the results as much as the attribute assignment uncertainty. The MCE fuzzy logic operator uncertainty affects the results the most. Confidence levels of 95% and 60% are evaluated with threshold values of 0.7 and 0.5 and show that accepting more uncertainty in the results increases the total area available for decision-making. Threshold values and confidence levels should be predetermined, although a series of combinations may yield the best decision-making support.