Shortening the duration of antibiotic therapy for patients admitted to hospital with community-acquired pneumonia should help reduce antibiotic consumption and thus bacterial resistance, adverse events, and related costs. We aimed to assess the need for an additional 5-day course of β-lactam therapy among patients with community-acquired pneumonia who were stable after 3 days of treatment. We did this double-blind, randomised, placebo-controlled, non-inferiority trial (the Pneumonia Short Treatment [PTC]) in 16 centres in France. Adult patients (aged ≥18 years) admitted to hospital with moderately severe community-acquired pneumonia (defined as patients admitted to a non-critical care unit) and who met prespecified clinical stability criteria after 3 days of treatment with β-lactam therapy were randomly assigned (1:1) to receive β-lactam therapy (oral amoxicillin 1 g plus clavulanate 125 mg three times a day) or matched placebo for 5 extra days. Randomisation was done using a web-based system with permuted blocks with random sizes and stratified by randomisation site and Pneumonia Severity Index score. Participants, clinicians, and study staff were masked to treatment allocation. The primary outcome was cure 15 days after first antibiotic intake, defined by apyrexia (temperature ≤37·8°C), resolution or improvement of respiratory symptoms, and no additional antibiotic treatment for any cause. A non-inferiority margin of 10 percentage points was chosen. The primary outcome was assessed in all patients who were randomly assigned and received any treatment (intention-to-treat [ITT] population) and in all patients who received their assigned treatment (per-protocol population). Safety was assessed in the ITT population. This study is registered with ClinicalTrials.gov, NCT01963442, and is now complete. Between Dec 19, 2013, and Feb 1, 2018, 706 patients were assessed for eligibility, and after 3 days of β-lactam treatment, 310 eligible patients were randomly assigned to receive either placebo (n=157) or β-lactam treatment (n=153). Seven patients withdrew consent before taking any study drug, five in the placebo group and two in the β-lactam group. In the ITT population, median age was 73·0 years (IQR 57·0-84·0) and 123 (41%) of 303 participants were female. In the ITT analysis, cure at day 15 occurred in 117 (77%) of 152 participants in the placebo group and 102 (68%) of 151 participants in the β-lactam group (between-group difference of 9·42%, 95% CI -0·38 to 20·04), indicating non-inferiority. In the per-protocol analysis, 113 (78%) of 145 participants in the placebo treatment group and 100 (68%) of 146 participants in the β-lactam treatment group were cured at day 15 (difference of 9·44% [95% CI -0·15 to 20·34]), indicating non-inferiority. Incidence of adverse events was similar between the treatment groups (22 [14%] of 152 in the placebo group and 29 [19%] of 151 in the β-lactam group). The most common adverse events were digestive disorders, reported in 17 (11%) of 152 patients in the placebo group and 28 (19%) of 151 patients in the β-lactam group. By day 30, three (2%) patients had died in the placebo group (one due to bacteraemia due to Staphylococcus aureus, one due to cardiogenic shock after acute pulmonary oedema, and one due to heart failure associated with acute renal failure) and two (1%) in the β-lactam group (due to pneumonia recurrence and possible acute pulmonary oedema). Among patients admitted to hospital with community-acquired pneumonia who met clinical stability criteria, discontinuing β-lactam treatment after 3 days was non-inferior to 8 days of treatment. These findings could allow substantial reduction of antibiotic consumption. French Ministry of Health.