The technical condition of multi-family residential buildings is one of the most important factors determining their aging process. It is most often determined based on the subjective assessment of the assessor without relying on unified criteria. The aim of the article is to propose a systematization of the method of determining the technical condition of buildings using machine learning algorithms. Based on a series of studies in residential buildings erected using traditional technology, a database was created, including characteristic features of individual buildings and their elements, occurring defects and renovation management conducted in individual years. The collected database was used to conduct an analysis using machine learning (ML) algorithms representing various predictive techniques, including Gradient Boosting, Random Forest and k-NN. Statistical analyses were performed for each of the algorithms to confirm the correctness of the estimated models. Variables that have a significant impact on the estimation process for individual algorithms were also determined. The use of machine learning allowed for a significant increase in the objectivity and precision of forecasting the technical condition of buildings. The developed models demonstrate practical potential. It is advisable to extend the research based on a larger database in the future, which may result in even greater accuracy in forecasting the technical condition of residential buildings.