This study explores how well artificial neural networks can predict the reduction in compressive strength of recycled aggregate concrete when exposed to different heat conditions. This study utilizes a dataset collected from previous experiments to train and validate the artificial neural network models. Two models are developed to estimate compressive strength based on several factors, including the mix proportions, heat conditions, physical properties of aggregates, and the age of the concrete. Results show that factors like maximum temperature, exposure time, recycled aggregate replacement ratio, and water-to-cement ratio have a significant impact on compressive strength, especially in high temperatures. The models demonstrated high accuracy, achieving about 97% correlation and low error rates, making them reliable for real-world use. Analyses to understand the interactions between the input factors and the compressive strength better are also conducted. The results indicate that compressive strength decreases more with longer exposure to high temperatures and that a water-to-cement ratio of 0.5 performs better than 0.4 at temperatures up to 100 °C. Moreover, a higher recycled aggregate replacement ratio leads to less strength loss up to 200 °C, but strength loss increases significantly beyond that temperature. This research promotes the use of recycled aggregate in structures vulnerable to high temperatures such as chimneys and helps structural engineers in evaluating repair strategies for concrete constructions post-elevated heat exposure.