OBJECTIVES: Due to setting specific characteristics the economic attractiveness of new technologies might vary between hospitals. We compared classical and statistical process control (SPC) methods for measuring and monitoring the real-world cost-effectiveness of new technologies in hospitals. METHODS: A systematic literature review was performed in PubMed in April 2009 to identify studies applying classical and SPC methods for investigating the (cost-)effectiveness of changes in inpatient health care processes. Both methodologies were compared using a predefined set of criteria such as accuracy, flexibility, informative value, suitability and user-friendliness. RESULTS: Classical statistical methods based on ‘time static’ (cross-sectional) statistical analysis with aggregated data are widely used. With the ability to detect statistical significant differences classical methods may provide higher accuracy. They are characterized by large one-time data collections to evaluate the impact of a process change for a pre-specified time period, limiting their flexibility. SPC methods which analyze time series data by monitoring a process over time have been used rarely but their application is increasing. They combine time series analysis with a graphical representation of the data. Patterns in time series data contain important information which other methods reliant on averages (or other summary statistics) could mask. By providing continuous feedback SPC is capable not only of detecting the results of process changes earlier but also of monitoring the process sustainability. SPC can be applied to routine data easier as it is typically less sensitive to statistical issues. Furthermore, the graphical representation of the data has advantages because statistical measures such as P-values are often poorly understood and misinterpreted. CONCLUSIONS: Both methodologies are suitable for measuring the (cost)-effectiveness of changes in health care processes. SPC seems to be the preferred methodology under real-world conditions to support decision-making although it commonly does not achieve the accuracy of classical statistical methods.