Objective: This article proposes a prediction model applicable to the propagation of noise generated by fixed sources as the result of an analysis of phenomena related to the generation and propagation of sound levels and the subsequent correlation between the levels estimated with the data recorded in the field. Materials and methods: An experimental program was designed that included the measurement of sound pressure levels with a sound level meter to different weather conditions and distances from the noise emission source to compare them with the levels estimated by ISO 9613 Part 2. A statistical analysis of the data recorded in field was performed to observe the dependence with the meteorological variables recorded during the measurements. Results and discussion: Standard error for the proposed prediction method is 11.4 dB, and absolute average error is 9.1 dB. The correlation coefficient of the proposed model is 0.87. A statistically significant relationship exists between the variables with a 95.0% confidence level. Conclusion: A propagation model was obtained that presented a better fit than the method of ISO 9613 Part 2 and a higher correlation coefficient.