OSLOM is a method for detecting statistically significant communities in networks, capable of handling edge directions, weights, overlapping communities, hierarchies, and dynamic structures. It optimizes the statistical significance of clusters using a fitness function based on extreme and order statistics. OSLOM can be used alone or as a refinement tool for other community detection methods. It has been tested on artificial benchmark graphs and real networks, showing comparable performance to existing algorithms. OSLOM is implemented in a freely available software (http://www.oslom.org) and is suitable for analyzing both static and dynamic networks. The method distinguishes between significant and pseudo-communities, handles overlapping communities, and detects hierarchical structures. It also accounts for weighted and directed graphs, and can analyze time-stamped networks to study community dynamics. OSLOM is efficient and scalable, with performance comparable to other methods on large networks. It has been tested on various benchmark graphs, including the LFR benchmark, showing strong performance in detecting communities, especially in cases with overlapping and hierarchical structures. OSLOM is able to handle random graphs and noise, identifying communities that are not randomly distributed. The method is robust and effective in identifying meaningful community structures in complex networks.OSLOM is a method for detecting statistically significant communities in networks, capable of handling edge directions, weights, overlapping communities, hierarchies, and dynamic structures. It optimizes the statistical significance of clusters using a fitness function based on extreme and order statistics. OSLOM can be used alone or as a refinement tool for other community detection methods. It has been tested on artificial benchmark graphs and real networks, showing comparable performance to existing algorithms. OSLOM is implemented in a freely available software (http://www.oslom.org) and is suitable for analyzing both static and dynamic networks. The method distinguishes between significant and pseudo-communities, handles overlapping communities, and detects hierarchical structures. It also accounts for weighted and directed graphs, and can analyze time-stamped networks to study community dynamics. OSLOM is efficient and scalable, with performance comparable to other methods on large networks. It has been tested on various benchmark graphs, including the LFR benchmark, showing strong performance in detecting communities, especially in cases with overlapping and hierarchical structures. OSLOM is able to handle random graphs and noise, identifying communities that are not randomly distributed. The method is robust and effective in identifying meaningful community structures in complex networks.