Purpose – Project cost is normally a key performance indicator for all projects, and therefore features prominently in benchmarking exercises aimed at identifying best practice. However, projects in different locations first require all costs to be expressed in equivalent units. Failing to do this leads to erroneous and unreliable results. The paper aims to discuss these issues. Design/methodology/approach – Applying international construction as the focus for the study, cost data from 23 cities worldwide are compared using a range of methods including currency conversion and purchasing power parity (PPP). Coefficient of variation (CoV) forms the test for identifying the method with the lowest volatility. Findings – It is found that purchasing power is the preferable theoretical base for international cost conversion, and currency conversion (frequently used by practitioners) is not recommended. The citiBLOC PPP method has the lowest CoV across the data set and therefore more closely reflects the Law of One Price that underpins the concept of PPP. Originality/value – This research highlights the importance of a valid cost conversion methodology to properly understand the comparative performance of projects. Its application to benchmarking is demonstrated using the data envelopment analysis method.