Conventional object-oriented programming languages like C "+"+ and Smalltalk aren't able to perform a consequent object-oriented parallel programming. The division into object is a natural partioning of the load. It's possible to use this way of automatic distribution for multi-processor sytems with distributed memory. On the base of the object-oriented parallel programming language OPAL, developed at the Technical University of Dresden, a tuning system was developed and tested. This tuning system is suitable to adapt object-oriented programs to parallel processing. The postprocessing of the monitoring data delivers facts to optimize the load distribution, the choice of a suitable processor topology and the source program. The tun...