While immune checkpoint inhibitors (ICIs) are increasingly reshaping the therapeutic landscape of non-small-cell lung cancer (NSCLC), only a limited proportion of patients achieve a relevant and long-lasting benefit with these treatments, calling for the identification of clinical and, ideally modifiable, predictors of efficacy. Body composition phenotypes may reflect aspects of patients' immunology and thereby their ability to respond to ICIs. This study aims to explore the possible association between pre-treatment body composition phenotypes, tumour response, and clinical outcomes in patients receiving first-line pembrolizumab monotherapy for advanced NSCLC. A retrospective review of consecutive patients with treatment-naïve NSCLC and PD-L1 expression ≥50% undergoing pembrolizumab at three academic institutions was performed. Pre-treatment body composition parameters were measured at the third lumbar vertebra level by computed tomography, defined using pre-established cut-offs. Primary endpoint was objective response rate (ORR), secondary endpoints progression-free survival and overall survival (PFS and OS), compared through the log-rank test and the Cox proportional hazards model. Data from 134 patients (93 males [69.4%] and 41 females [30.6%]) were collected. Median age was 69years (range 36-85), with a median follow-up of 12months (range 1-131). The median body mass index (BMI) was 24.5 (IQR 21.5; 26.1) kg/m2. Overall, 59.0% and 51.5% of patients met established radiographic criteria for evidence of sarcopenia and myosteatosis, respectively, which occur across the BMI spectrum. Multivariate regression analysis, adjusted for co-morbidities, revealed that sarcopenia (aOR 5.56, 95% CI. 2.46-12.6, P<0.0001) and low intermuscular adipose tissue (IMAT) area (aOR 1.83, 95% CI. 1.22-2.83, P=0.001) were associated with a lower rate of ORR (30.4% vs. 70.5%, P<0.0001 and 30.7% vs. 73.2%, P<0.0001, respectively). Moreover, both in univariate and multivariate analysis, adjusted for co-morbidities, low performance status according to the Eastern Cooperative Oncology Group scale (ECOG PS), sarcopenia and low IMAT were significantly related to short PFS (ECOG PS: aHR 2.73, 95% CI 1.60-4.66, P<0.0001; sarcopenia: aHR 2.24, 95% CI 1.37-3.67, P=0.001; IMAT depot: aHR 2.26, 95% 1.40-3.63, P=0.002) and OS (ECOG PS: aHR 3.44, 95% CI 1.96-6.01, P<0.0001; sarcopenia: aHR 4.68, 95% CI 2.44-8.99, P<0.0001; IMAT depot: aHR 3.18, 95% 1.72-5.88, P<0.0001). Skeletal muscle abnormalities, apparently frequent in NSCLC, potentially represent intriguing predictive markers of response to ICIs and survival outcomes. Large prospective trials are needed to validate ICIs responders' clinical biomarkers.