Precipitation and deposition of asphaltene during petroleum production is a challenging problem confronted by the oil industry compromising the profitability of production fields through loss efficiency of recovery process as well as create remedial cost. Hence, developing a robust model for assessment of asphaltene stability in crude oil is necessary. Delta RI (Delta RI = RI - PRI) is a novel criterion for stability determination of asphaltene in crudes. An integrated intelligent method, called neuro-fuzzy (NF) has been used in this study for estimation of Delta RI from SARA fraction data. NF develops a fuzzy inference system which is subsequently optimized by virtue of learning capability of neural network (NN). Since NN structure, embedded in NF systems is highly at risk of sticking in local minima, another improved fuzzy model is constructed and is subsequently optimized by virtue of particle swarm optimization (PSO) technique. Correlation coefficients for neuro-fuzzy and PSO-fuzzy model are found to be 0.857 and 0.9102, respectively. Comparison between constructed models show optimization of fuzzy model by virtue of PSO technique significantly improves accuracy of final prediction. Implementation of the proposed method indicate that PSO-fuzzy model is capable of accurately predicting asphaltene stability.