This study assessed the between- and within-match variability of external training load measures during two consecutive Twenty20 cricket seasons in professional fast bowlers. Global positioning system (GPS) and accelerometer data (PlayerLoad™) were collected from eight fast bowlers in 17 matches of domestic Twenty20 competition. Using GPS-accelerometry systems the variables selected for analysis were; total distance, low speed running distance (≤14.4 km.h-1), high-speed running distance (≥14.4 km.h-1), total sprint distance (≥18 km.h-1), number of sprint efforts (n), peak speed (km.h-1) and PlayerLoad™ (arbitrary units; AU). These variables were further categorised into specific reference periods; between-match (overall and bowling only) and within-match (between-over). Data were log transformed and the coefficient of variation (CV) and between-subject standard deviation determined (both expressed as percentages). The data shows that between-match variability was greatest in high-speed running distance (32.9% CV), total sprint distance (49.0% CV) and number of sprint efforts (48.0% CV). Similarly, within-match between-over data was greatest in high-speed running distance (12.8% CV), total sprint distance (17.1% CV) and number of sprint efforts (12.3% CV), yet this variability was markedly reduced compared to between-match observations. The results show that global measures of external training load (total distance and PlayerLoad™; 5.5-13.3% CV) are relatively stable, yet high-speed locomotive activities exhibit a larger degree of variability both between- and within-match. These findings have importance for practitioners, who seek to facilitate performance by informed training prescription based on replicating match demands.