Early and adequate empiric antibiotic therapy is essential in the treatment of pneumonia–sepsis, influencing clinical outcomes. This retrospective before–after study aimed to appraise the impact of local ASP (written guidelines and antibiotic restriction) on antibiotic (AB) use and clinical outcomes in patients requiring intensive care due to pneumonia–sepsis. This single-center retrospective observational study was conducted in the ICU of a pulmonology department in Hungary. Data collection for the pre-intervention period occurred from January 2018 to May 2022, while that for the ASP period was from June 2022 to March 2024. The patients admitted to the ICU with pneumonia–sepsis were mainly men (58/101, 57.4% and 84/128, 65.6%); the need for ICU increased with age, and most of the patients belonged to 65+ age group in both study phases (68.3% and 58.6%). The majority of the patients had four or more comorbidities (57.4% and 40.6%). In-hospital mortality was relatively high (42.6% and 41.4%), with most of the patients already losing their lives in the ICU (33/43, 76.7% and 37/53, 69.8%). A significant increase in guideline-adherent agent selection (34.5%) and the use of combination therapy (35.0%) was observed, while the use of fluoroquinolones decreased significantly (-31.1%). In the after period, a significant decrease in the number of patients using restricted ABs (-53.3%) was observed. In one-third of these cases (10/34, 29.4% and 16/40, 40%), 2–4 MDR pathogens were detected simultaneously, resulting in a significant increase in direct costs (10.5%). The use of inappropriate AB therapy was relatively low in the presence of MDRs in both phases (5.9% and 15%). In the ASP period, guideline adherence was associated with slightly better clinical outcomes (30-day mortality: -0.8% and LOS: -22.6%). Interrupted time series analysis (ITS) indicates a decrease in the trend of AB use in the ICU (from 0.16 /30 days to -0.22 /30 days). ASP implementation in the ICU resulted in a significant improvement in the appropriate use of ABs, and guideline adherence led to slightly better clinical outcomes. ASP may hold promise in improving AMR with a sustained long-term effect.