This study quantified average and peak external intensities of various basketball training drills. Thirteen youth male basketball players (age: 15.2 ± 0.3 years) were monitored (BioHarness-3 devices) to obtain average and peak external load per minute (EL · min; peak EL · min) during team-based training sessions. Researchers coded the training sessions by analysing the drill type (skills, 1vs1, 2vs2, 3vs0, 3vs3, 4vs0, 4vs4, 5vs5, 5vs5-scrimmage), court area per player, player's involvement in the drill (in percentage), playing positions (backcourt; frontcourt) and competition rotation status (starter; rotation; bench). Separate linear mixed models were run to assess the influence of training and individual constraints on average and peak EL · min. Drill type influenced average and peak EL · min (p < 0.05), but with different directions of effects. EL · min was higher in skills and 4vs0 drills, while higher peak EL · min values were obtained in 5vs5 and 5vs5-scrimmage. Similarly, EL · min was higher when involvement % increased (p = 0.001), while there was an opposite trend for peak EL · min (lower with higher involvement %). Court area per player influenced peak (p = 0.025) but not average demands. No effects were found for playing position or competition rotation status (all p > 0.05), except for a moderately higher EL · min in starters compared to bench players. The external load intensities of basketball training drills substantially vary depending on the load indicator chosen, the training content, and task and individual constraints. Practitioners should not interchangeably use average and peak external intensity indicators to design training but considering them as separate constructs could help to gain a better understanding of basketball training and competition demands.