A topological-mathematical model obtained by linear discriminant analysis has been used to the search of new nonsteroidal antinflammatory drugs (NSAIDs). After carrying out an in silico screening based on such a model, on the Aldrich database, new structures potentially active were selected. Among these structures stand fourteen compounds, from which only one had been previously recorded as NSAID in the literature. The experimental tests performed on the remaining substances demonstrated that several compounds showed either in vitro or in vivo or both activity. Moreover, four compounds, namely 1,3-bis(benzyloxycarbonyl)-2-methyl-2-thiopseudourea, 4,6-dichloro-2-methylthio-5-phenylpyrimidine, 2-chloro-2,6-acetoxylidide and trans-1,3-diphenyl-2-propen-1-ol, showed a significant in vivo antinflammatory activity as compared to the reference drug (indomethacin). These results reinforce the role of Molecular Topology as a useful tool for drug discovery.