•A new AO classification for trochanteric fractures was recently published; no studies have evaluated its inter- and intra-observer agreement.•We assessed 68 intertrochanteric fractures; fractures were classified using the original and the new AO classifications.•The inter-observer agreement using the new AO classification was significantly better than using its original version.•The new AO classification system allowed better agreement when distinguishing stable from unstable patterns. A new AO classification for intertrochanteric fractures was recently published; no studies have evaluated its inter- and intra-observer agreement. Six evaluators (three hip subspecialists and three residents) assessed radiographs of 68 intertrochanteric fractures; fractures were classified using the original and the new AO classifications. The cases were displayed in a random sequence after a six-week interval for repeat evaluation. We used the Kappa coefficient (k) to determine inter- and intra-observer agreement. Inter-observer agreement was slight (k = 0.128 0.092–0.170) using the original and fair (k = 0.250 0.186–0.327), with the new AO classification. Orthopedic residents exhibited better agreement than hip surgeons using the original classification (k = 0.302 0.210–0.416 and k= -0.018 -0.058–0.029, respectively) and the new classification (k = 0.388 0.294–0.514 and k = 0.109 0.031–0.192, respectively). Using both classifications as dichotomous variables (stable or unstable patterns), the agreement was slight (k = 0.158 0.074–0.246) using the original classification and moderate (k = 0.425 0.308–0.550) with the new AO classification. The agreement was fair using the original (k = 0.350 0.278–0.424) and the new (k = 0.295 0.239 to 0.353) AO classifications, respectively. Residents had better agreement than hip specialists using the original (k = 0.405 0.303–0.512) versus (k = 0.292 0.193–0.293) and the new classification (k = 0.449 0.370 to 0.528 versus k = 0.129 0.064 to 0.208). The inter-observer agreement using the new AO classification was significantly better than using its original version. Also, the new AO classification system allowed better agreement when distinguishing stable from unstable patterns.