To assess the impact of a long-term initiative to improve safety culture among professionals working in a Health Area, and to know their perceived usefulness.An uncontrolled intervention study was designed in a public health care organization including a 3rd level hospital and 5,000 professionals. To measure the impact, the AHRQ Survey was conducted by telephone. A total of 7 dimensions of culture were measured, before starting the project (2012, n=100) and 3 years later (2015, n=207). Variations between 2012 and the respondents aware of the project in 2015 (RAP) were compared, as also between this last group and the rest of respondents (RNAP). The utility was assessed using a 5-item Likert scale, defining higher utility by medians 4 or higher.The response rates were above 80%. In 2015, the 41.5% of respondents were RAP (95%CI: 34.8-48.3), which was perceived as of high utility. Negative variations were detected in "sense of security" (-9.9%, P<.01, vs. 2012, and -4.2% between 2015 groups) and "feedback and communication errors" (-10.0% vs. 2012, and -8.9% between 2015 groups, P<.05). There was a not-significant positive variation in "openness in communication" (1.3% vs. 2012, and 6.9% between 2015 groups). The "management support" showed a not-significant improve in 2015 (37.0%, 95%CI: 30.9-43.1, in RAP; and 38.3%, 95%CI: 33.1-43.4, in RANP) in comparison to 2012 (31.4%, 95%CI: 28.4-39.7).A paradoxical worsening is detected in several dimensions, this probably due to immaturity of the organization and the instrument used. Thus, tools explicitly considering the degree of maturity may be more appropriate to measure cultural changes, although more studies are needed.