The definition and modeling of business processes may vary from one context to another, for example, to adapt to new business requirements, to regulations in different regions or to reflect new resource allocations. These differences often lead to the definition of several variants of the same process and can be reflected in different process perspectives such as control-flow, data, resources or perfor- mance. The management of process variants can be a laborious, time-consuming and error-prone task since they require a high coordination in the management of each variant and in most cases this management is done manually. Many pro- posals have been developed to deal with the variability of business processes. However, none of them covers in detail the variability in the performance per- spective, which is concerned with the definition of performance requirements usually specified as a set of Process Performance Indicators (PPIs). This vari- ability can be reflected in the form of repetitive and redundant PPI definitions, and can lead to errors and inconsistencies in PPI definitions. To address this problem, we propose a detailed PPI variability classification and a formalization on how PPIs can be modeled together with the variability of other process per- spectives. To this end, we considered variability management approaches, called by restriction and by extension, and we illustrated our proposal by integrating it with two existing process variability modeling languages. An evaluation con- ducted in two scenarios shows the feasibility of our approach and how it can be successfully used to model the variability that is present in those scenarios.