The concept of model-based performance monitoring has been tested in a recovery boiler operating in a Swedish kraft pulp mill. Key performance indicators (KPI) have been calculated and used to visualise the degree of fouling of boiler bank, super heaters and economiser of the recovery boiler. The KPIs illustrate the actual degree of fouling independently from the actual thermal load. The presented tool enables an early warning for massive fouling in the boiler. Even smaller changes in boiler performance can be followed enabling operation close to maximum capacity. The tool may also support operators to test modifications of the operation, for instance. A trend for the sticky temperature of dust particles has been added as a first approach to provide a measure for a critical flue gas temperature. That trend in combination with an indirectly measured flue gas temperature adds further value to model-based performance monitoring.