The method, which uses the arithmetic mean and standard deviation (M₁) of total heights (Ht), is often used in the stratification of uneven-aged forests and assumes that the Ht must follow a normal standard curve considering a standard deviation of the mean. In cases where this assumption is rejected, alternative methods are used, such as the fundamentals in multivariate analysis (M₂) and Lorey’s mean height (M₃). The objective was to evaluate the method M1 considering its assumption and comparing it with methods M₂ and M₃. The data were collected in 308 contiguous plots of 100 m², installed in an Alluvial Ombrophylous Forest. The assumption of the method M₁ was analyzed adjusting a Normal Probabilistic Density Function, evaluated by the K-S test. The methods were evaluated using discriminant analyses, structural analyses and three-dimensional vertical profiles. It was verified that the assumption was not met. The discriminant analyses showed that the classification was 100 % correct in methods M₁ and M₃. The 13 most important species, in descending order, were the same in all three methods. However, from the 13^th, the differences were more remarkable in the method M₁. Firstly, it is recommended to use the method M₃, as it is more practical and provides results that are equal to those obtained using the method M₂. However, if the objective is to obtain a higher level of detail, the option is the method M₂ since it allows obtaining up to five strata with 100 % correct classification.