The main objective of this study was to develop, implement, and evaluate a set of forecasting models based on mathematical relations for predicting daily minimum and maximum air temperatures in Guilan Province. For this purpose, daily meteorological data from ten selected weather stations across the province were collected over the period 2011–2021. Following data quality control procedures, correlation analysis was conducted among various atmospheric parameters to identify the most influential variables affecting temperature variations. Based on the results of this analysis, a total of eleven mathematical models, including linear, polynomial, exponential, power, and logarithmic formulations, were developed separately for forecasting daily minimum and maximum temperatures. Subsequently, the predictive performance of the proposed mathematical models was evaluated and compared with that of artificial neural network (ANN) models. The results indicated that among the considered meteorological variables, the temperature of the previous day exhibited the strongest dependency and the greatest influence on the temperature of the following day. Furthermore, comparison of the evaluation metrics demonstrated that the mathematical models and artificial neural networks achieved relatively similar and acceptable accuracy in short-term temperature forecasting. Considering their structural simplicity, transparency of relations, lower data requirements, and stable performance, mathematical models can be regarded as efficient and reliable tools for short-term temperature prediction. Nevertheless, extending the applicability of the findings to long-term forecasting horizons or to regions with different climatic conditions requires further comprehensive investigations and additional analyses.