(1) Despite the effectiveness of immune checkpoint inhibitors (ICIs) in lung cancer, there is a lack of knowledge about predictive biomarkers. The objective of our study is to analyze different subsets of T-lymphocytes and natural killer (NK) cells as predictive biomarkers in a cohort of patients with nonsmall cell lung cancer (NSCLC) treated with ICI. (2) This is an observational, prospective study with 55 NSCLC patients treated with ICI. A total of 43 T and NK cell subsets are analyzed in peripheral blood, including the main markers of exhaustion, differentiation, memory, activation, and inhibition. (3) Regarding the descriptive data, Granzyme B+CD4+ Treg lymphocytes stand out (median 17.4%), and within the NK populations, most patients presented cytotoxic NK cells (CD56+CD3-CD16+GranzymeB+; median 94.8%), and about half of them have highly differentiated adaptive-like NK cells (CD56+CD3-CD16+CD57+ (mean 59.8%). A statistically significant difference was observed between the expression of PD1 within the CD56bright NK cell subpopulation (CD56+CD3-CD16-PD-1+) (p = 0.047) and a better OS. (4) Circulating immune cell subpopulations are promising prognostic biomarkers for ICI. Pending on validation with a larger sample, here we provide an analysis of the major circulating T and NK cell subsets involved in cancer immunity, with promising results despite a small sample size.