There are regions in the state of Rio Grande do Sul where rainfall is sufficient for agricultural functions and social maintenance, but in other locations this occurrence is not so satisfactory that water supply is sufficient for the basic needs of the population. In this sense, knowing how to accurately estimate weather variations is of great importance. Statistical modeling of extreme levels is associated with probability distributions, belonging to a class of distributions in a branch of statistics called Extreme Value Theory (EVT). The objective of this work was to analyze the occurrence of maximum rainfall in the city of Santa Maria - RS through the Generalized Extreme Values, Gumbel and Pearson type III distributions and to estimate the amount of rainfall for return periods of 2, 5, 10, 15, 30, 50 and 100 years. The results of the KS test and Histogram plot plots indicate that the three probability distributions fit the data and, between the Gumbel and GEV distributions, the likelihood ratio test indicated that the Gumbel distribution is more appropriate. However, the mean absolute percentage error revealed that the Pearson III distribution provided more accurate estimated return levels for some months.Existem regiões no estado do Rio Grande do Sul onde ocorrência de chuva é suficiente para as funções agrícolas e manutenção social, porém em outras localidades esta ocorrência não é tão satisfatória a ponto de o fornecimento d’água ser suficiente para as necessidades básicas da população. Nesse sentido, estimar com certa precisão as variações meteorológicas é de grande importância, uma vez que viabiliza um planejamento mais preciso nestas regiões. O objetivo desse trabalho foi analisar a ocorrência de precipitações pluviais máximas na cidade de Santa Maria – RS por meio das distribuições Generalizada de Valores Extremos, Gumbel e Pearson tipo III e estimar a quantidade de chuva para períodos de retorno de 2, 5, 10, 15, 30, 50 e 100 anos. Os resultados do teste de Kolmogorov Kolmogorov-Smirnov e os Histogramas com as funções densidades de probabilidades ajustadas indicam que as três distribuições de probabilidade se ajustaram aos dados e, entre as distribuições Gumbel e GVE, o teste de razão de verossimilhanças apontou que a distribuição Gumbel é mais adequada. Contudo o erro percentual absoluto médio revelou que a distribuição Pearson III forneceu níveis de retorno estimados mais precisos em alguns meses. Palavras-chave: Teoria de valores extremos; probabilidade de ocorrência; Distribuição Generalizada de Valores Extremos (GEV); Distribuição Gumbel; Distribuição de Pearson Tipo III; climatologia.   Maximum monthly rainfall in Santa Maria, RS   ABSTRACT: There are regions in the state of Rio Grande do Sul where rainfall is sufficient for agricultural functions and social maintenance, but in other locations this occurrence is not so satisfactory that water supply is sufficient for the basic needs of the population. In this sense, knowing how to accurately estimate weather variations is of great importance. Statistical modeling of extreme levels is associated with probability distributions, belonging to a class of distributions in a branch of statistics called Extreme Value Theory (EVT). The objective of this work was to analyze the occurrence of maximum rainfall in the city of Santa Maria - RS through the Generalized Extreme Values, Gumbel and Pearson type III distributions and to estimate the amount of rainfall for return periods of 2, 5, 10, 15, 30, 50 and 100 years. The results of the KS test and Histogram plots indicate that the three probability distributions fit the data and, between the Gumbel and GEV distributions, the likelihood ratio test indicated that the Gumbel distribution is more appropriate. However, the mean absolute percentage error revealed that the Pearson III distribution provided more accurate estimated return levels for some months. Keywords: Extreme value theory; probability of occurrence; Generalized Distribution of Extreme Values (GEV); Gumbel Distribution; Pearson Distribution Type III; climatology.