Background: Surrogate biomarkers of efficacy are needed for anti-PD1/PD-L1 therapy, given the existence of delayed responses and pseudo-progressions. We evaluated changes in serum IL-8 levels as a biomarker of response to anti-PD-1 blockade in melanoma and non-small-cell lung cancer (NSCLC) patients. Patients and methods: Metastatic melanoma and NSCLC patients treated with nivolumab or pembrolizumab alone or nivolumab plus ipilimumab were studied. Serum was collected at baseline; at 2-4 weeks after the first dose; and at the time-points of response evaluation. Serum IL-8 levels were determined by sandwich ELISA. Changes in serum IL-8 levels were compared with the Wilcoxon test and their strength of association with response was assessed with the Mann-Whitney test. Accuracy of changes in IL-8 levels to predict response was estimated using receiver operation characteristics curves. Results: Twenty-nine melanoma patients treated with nivolumab or pembrolizumab were studied. In responding patients, serum IL-8 levels significantly decreased between baseline and best response (P < 0.001), and significantly increased upon progression (P = 0.004). In non-responders, IL-8 levels significantly increased between baseline and progression (P = 0.013). Early changes in serum IL-8 levels (2-4 weeks after treatment initiation) were strongly associated with response (P < 0.001). These observations were validated in 19 NSCLC patients treated with nivolumab or pembrolizumab (P = 0.001), and in 15 melanoma patients treated with nivolumab plus ipilimumab (P < 0.001). Early decreases in serum IL-8 levels were associated with longer overall survival in melanoma (P = 0.001) and NSCLC (P = 0.015) patients. Serum IL-8 levels also correctly reflected true response in three cancer patients presenting pseudoprogression. Conclusions: Changes in serum IL-8 levels could be used to monitor and predict clinical benefit from immune checkpoint blockade in melanoma and NSCLC patients.