Linezolid is a last-resort antibiotic for infections caused by multidrug-resistant microorganisms. It is widely used for off-label indications and for longer than recommended treatment durations, exposing patients at higher risk of adverse drug reactions (ADRs), notably thrombocytopenia. This study aimed to investigate ADR incidence and risk factors, identify thrombocytopenia-related trough levels based on treatment duration, and evaluate the performance of predictive scores for ADR development. Adult in-and outpatients undergoing linezolid therapy were enrolled in three hospitals and ADRs and linezolid trough levels prospectively monitored over time. A population pharmacokinetic (pop-PK model) was used to estimate trough levels for blood samples collected at varying times. A multivariate analysis based on 63 treatments identified treatment duration ≥ 10 days and trough levels > 8 mg/L as independent risk factors of developing thrombocytopenia, with high trough values correlated with impaired renal function. Five patients treated for > 28 days did not develop thrombocytopenia but maintained trough values in the target range (< 8 mg/L). The Buzelé predictive score, which combines an age-adjusted Charlson comorbidity index with treatment duration, demonstrated 77% specificity and 67% sensitivity to predict the risk of ADR. Thus, our work supports the necessity of establishing guidelines for dose adjustment in patients with renal insufficiency and the systematic use of TDM in patients at-risk in order to keep trough values ≤ 8 mg/L. The Buzelé predictive score (if ≥ 7) may help to detect these at-risk patients, and pop-PK models can estimate trough levels based on plasma samples collected at varying times, reducing the logistical burden of TDM in clinical practice.