A mixed effects model of evolution (MEME) was developed to detect individual sites subject to episodic diversifying selection. Unlike traditional models that assume constant selective pressure across all lineages, MEME allows for variation in selection pressure at the site and branch levels. This approach enables the identification of sites where positive selection acts on only a subset of lineages, which is often overlooked by existing methods. Using empirical and simulated data, MEME outperformed older models in detecting episodic positive selection. The study found that episodic selection is widespread, and the number of sites experiencing positive selection may have been vastly underestimated. MEME was tested on diverse datasets and showed superior performance in identifying sites under episodic selection compared to fixed effects models. It also demonstrated the ability to detect specific branches under diversifying selection, even when other branches were under purifying selection. The results suggest that natural selection is predominantly episodic, with transient periods of adaptive evolution masked by purifying or neutral selection on other branches. MEME provides a more accurate and powerful method for detecting selection at individual sites, particularly when selection is not uniform across lineages. The study highlights the importance of considering variable selection pressures in evolutionary analysis and underscores the limitations of traditional models that assume constant selective pressure. Overall, MEME offers a significant advancement in the detection of episodic diversifying selection, with implications for understanding the mechanisms of adaptive evolution in biological systems.A mixed effects model of evolution (MEME) was developed to detect individual sites subject to episodic diversifying selection. Unlike traditional models that assume constant selective pressure across all lineages, MEME allows for variation in selection pressure at the site and branch levels. This approach enables the identification of sites where positive selection acts on only a subset of lineages, which is often overlooked by existing methods. Using empirical and simulated data, MEME outperformed older models in detecting episodic positive selection. The study found that episodic selection is widespread, and the number of sites experiencing positive selection may have been vastly underestimated. MEME was tested on diverse datasets and showed superior performance in identifying sites under episodic selection compared to fixed effects models. It also demonstrated the ability to detect specific branches under diversifying selection, even when other branches were under purifying selection. The results suggest that natural selection is predominantly episodic, with transient periods of adaptive evolution masked by purifying or neutral selection on other branches. MEME provides a more accurate and powerful method for detecting selection at individual sites, particularly when selection is not uniform across lineages. The study highlights the importance of considering variable selection pressures in evolutionary analysis and underscores the limitations of traditional models that assume constant selective pressure. Overall, MEME offers a significant advancement in the detection of episodic diversifying selection, with implications for understanding the mechanisms of adaptive evolution in biological systems.