Abstract Objectives PTA of protein-unbound ceftriaxone may be compromised in critically ill patients with community-acquired pneumonia (CAP) with augmented renal clearance (ARC). We aimed to determine an optimized ceftriaxone dosage regimen based on the probability of developing ARC on the next day (PARC,d+1; www.arcpredictor.com). Patients and methods Thirty-three patients enrolled in a prospective cohort study were admitted to the ICU with severe CAP and treated with ceftriaxone 2 g once daily. Patients contributed 259 total ceftriaxone concentrations, collected during 1 or 2 days (±7 samples/day). Unbound fractions of ceftriaxone were determined in all peak and trough samples (n = 76). Population pharmacokinetic modelling and simulation were performed using NONMEM7.4. Target attainment was defined as an unbound ceftriaxone concentration >4 mg/L throughout the dosing interval. Results A two-compartment population pharmacokinetic model described the data well. The maximal protein-bound ceftriaxone concentration decreased with lower serum albumin. Ceftriaxone clearance increased with body weight and PARC,d+1 determined on the previous day. A high PARC,d+1 was identified as a clinically relevant predictor for underexposure on the next day (area under the receiver operating characteristics curve 0.77). Body weight had a weak predictive value and was therefore considered clinically irrelevant. Serum albumin had no predictive value. An optimal PARC,d+1 threshold of 5.7% was identified (sensitivity 73%, specificity 69%). Stratified once- or twice-daily 2 g dosing when below or above the 5.7% PARC,d+1 cut-off, respectively, was predicted to result in 81% PTA compared with 47% PTA under population-level once-daily 2 g dosing. Conclusions Critically ill patients with CAP with a high PARC,d+1 may benefit from twice-daily 2 g ceftriaxone dosing for achieving adequate exposure on the next day.