Objectives PD‐L1 immunohistochemistry (IHC) is an essential predictive biomarker for patients with non‐small cell lung cancer (NSCLC), required to inform treatment decisions regarding anti‐PD‐1 immune checkpoint inhibitor therapy. This study aims to investigate the concordance between PD‐L1 IHC assessed on NSCLC cytology and histology specimens and to determine the impactce of tumour cellularity. Methods Matched cytology and histology NSCLC specimens were retrieved from the archives of the Royal Melbourne Hospital and the Royal Prince Alfred Hospital. PD‐L1 IHC was performed concurrently on both specimens at the Peter MacCallum Cancer Centre using the SP263 assay kit on the Ventana Benchmark Ultra staining platform and scored by two experienced pathologists. Results Overall agreement between matched cytology and histology specimens was good (intraclass correlation coefficient = 0.653, n = 58); however, markedly increased when the analysis was limited to cell‐blocks with >100 tumour cells (intraclass correlation coefficient = 0.957, n = 29). Specificity at both 1% and 50% cut‐offs was high regardless of cellularity; however, sensitivity decreased in samples with <100 tumour cells. Conclusions PD‐L1 IHC on cytology cell‐block specimens in NSCLC is an acceptable alternative to histological specimens, provided adequate tumour cells are present. Clinicians and pathologists should be mindful of the risk of false negative PD‐L1 IHC in samples with low tumour cellularity, to avoid excluding patients from potentially beneficial treatment. This paper demonstrates that cytology cell blocks are acceptable specimens to assess PD‐L1 by immunohistochemistry in non‐small cell lung cancer, provided adequate tumour cells are present. Assessing insufficiently cellular samples risks a rate of false negatives, which may exclude patients from potentially beneficial treatment.