There is evidence that delivery room resuscitation of very preterm infants often deviates from internationally recommended guidelines. There were no published data in Spain regarding the quality of neonatal resuscitation. Therefore, we decided to evaluate resuscitation team adherence to neonatal resuscitation guidelines after birth in very preterm infants. We conducted an observational study. We video recorded resuscitations of preterm infants < 32 weeks’ gestational age and evaluated every step during resuscitation according to a score-sheet specifically designed for this purpose, following Carbine’s method, where higher scores indicated that more intense resuscitation maneuvers were required. We divided the score achieved by the total possible points per patient to obtain the percentage of adherence to the algorithm. We also compared resuscitations performed by staff neonatologists to those performed by pediatricians on-call. We compared percentages of adherence to the algorithm with the Chi-square test for large groups and Fisher’s exact test for smaller groups. We compared assigned Apgar scores with those given after analyzing the recordings and described them by their median and interquartile range. We measured the interrater agreement between Apgar scores with Cohen’s kappa coefficient. Linear and logarithmic regressions were drawn to characterize the pattern of algorithm adherence. Statistical analysis was performed using SPSS V.20. A p-value < 0.05 was considered significant. Our Hospital Ethics Committee approved this project, and we obtained parental written consent beforehand. Sixteen percent of our resuscitations followed the algorithm. The number of mistakes per resuscitation was low. Global adherence to the algorithm was 80.9%. Ventilation and surfactant administration were performed best, whereas preparation and initial steps were done with worse adherence to the algorithm. Intubation required, on average, 2.2 attempts; success on the first attempt happened in 33.3% of cases. Only 12.5% of intubations were achieved within the allotted 30 s. Many errors were attributable to timing. Resuscitations led by pediatricians on-call were performed as correctly as those by staff neonatologists. Resuscitation often deviates from the internationally recognized algorithm. Perfectly performed resuscitations are infrequent, although global adherence to the algorithm is high. Neonatologists and pediatricians need intubation training.