To carry out the research, was relevant to inquire about personnel quality services offered by Agencia Fiscal del Estado, through its tax payer’s attention module. Through SERVQUAL method, it is intended to measure the service quality offered in one of the most representative five branches in the city of Hermosillo, Sonora in relation to the size and the number of tax payers. The research took place during the last two weeks of June 2013 (June 17 to June 28, 2013). This investigation is framed within the quantitative approach with the procedure for the inductive-deductive knowledge acquisition. It is limited as an empiric investigation, beginning from an exploratory, descriptive level through a non-experimental-transversal design, because the information is handled in a qualitative manner for a better interpretation, but the original difference of opinion of the primary information sources are not modified, whereas phenomenon and positions were observed in a period of time of two weeks, from Monday to Friday, and the opinion of clients that came up to Agencia Fiscal to pay their contributions were taken for registration and subsequent analysis. (Kerlinger & Lee, 2002). In the tax collection offices the following clients’ unconformities about quality service are present: lack of interest in helping the client; slow service; lack of personnel; malfunctioning of the data base system; little empathy from personnel to clients; lack of knowledge about quality service among the personnel. For this reason, in this research this topic is taken to get better results and establish clients’ quality satisfaction level and the Service Quality Indicators (SQIs). Therefore, the following question is brought out: which is the clients’ satisfaction level and the service quality indicators in Agencia Fiscal del Estado located in the city of Hermosillo, Sonora? Regarding the size of the clients sample and with the purpose of generalizing the results, a sample was chosen according to the criteria for a probability sampling and for an infinite population (Hurtado, 2000), that emitted a sample of 96 people or clients for each one of the branches; however, for the convenience of the investigator, the final sample was of 100 clients, with a confidence coefficient of 95% and an error estimation of 10%, as is shown in the following Formula: