Monitoring agricultural drought utilizing the least variables can be useful at the areas with just rain gauge stations, especially in Iran, which is located in arid belt of the world. In this study, all of the stations having long duration data in Iran (31 stations with more than 59 years' data), have been used to calculate three drought indices: (1) Palmer Drought Severity Index (PDSI); (2) Standardized Precipitation Index (SPI); (3) Multivariate Standardized Precipitation Index (MSPI). According to the recommendation of world meteorological organization (WMO), Palmer index has been defined as the reference index and SPI9 (9-month time window of SPI), SPI10, SPI11 & SPI12, and also MSPI9-12 (9-12 months' time window), was evaluated by Palmer index. The results showed significant relation between SPI11 or SPI12 and PDSI in Iran's climates. MSPI9-12 has a significant relation with PDSI too, and the correlation of MSPI with Palmer Index was found to be stronger than the SPI, in all stations. The relation between MSPI and PDSI is also stronger in extra-arid climates, and weaker in humid and per-humid areas. So, MSPI can be used for agricultural drought monitoring in the places where there are restrictions in meteorological dataset (when precipitation data are only available). In the second part, prediction of MSPI9-12 has been done using Adaptive Neuro-Fuzzy Inference System (ANFIS) merged with bio-inspired optimization algorithms. The meta-heuristic models implemented are ANFIS-ACO (ANFIS merged with Ant Colony Optimization), ANFIS-GA (ANFIS merged with Genetic Algorithm) and ANFIS-PSO (ANFIS merged with Particle Swarm Optimization). The models had their best predictions in arid, semi-arid and Mediterranean climates while humid climate provided the weakest predictions. Among the mentioned algorithms, the ACO and GA had the best performances to optimize ANFIS; they improved the ANFIS's accuracy by 45.9% and 43.2%, respectively.