Introduction: Power meter is a useful tool for monitoring cyclists' training and race performance. However, limited data are available regarding BMX racing power output. The aim of this study was to characterise the power production of BMX riders and investigate its potential role on race performance. Methods: Fourteen male riders (age: 20.3 ± 1.5 years, height: 1.75 ± 0.05 m, mass: 70.2 ± 6.4 kg) participated in this study. The tests consist of performing two races apart from 15-min recovery. SRM power meter was used to record power and cadence. Cyclists' fastest race was used for the data analysis. Heart rate was recorded at 1-s intervals using a Garmin HR chest strap. Lap time was recorded using four pairs of photocells positioned at the start gate, bottom of the start ramp, end of first corner (time cornering), and on the finish line. Results: There was a large correlation between race time and relative peak power (r = −0.68, p < 0.01) as well as average power with zero value excluded (r = −0.52, p < 0.01). Race time was also significantly associated with time cornering (r = 0.58, p < 0.01). Peak power (1288.7 ± 62.6 W) was reached in the first 2.34 second of the race. With zero values included, the average power was 355.8 ± 25.4 W, which was about 28% of the peak power, compared to 62% when zero values were excluded (795.6 ± 63.5 W). Conclusion: The post-race analysis of the power data might help the cyclists recognizing the need to apply certain strategies on pedalling rates and power production in certain portions of the BMX track, specially, at the start and around the first corner. BMX coaches must consider designing training programs based on the race intensity and power output zones.