Your specific rules have so much more impact on these stats than any of the factors you mention such as numbers of rules and facts. The number of rules by itself has very little impact on performance, except at start-up time. There are knowledge bases out there with > 100,000 rules, which achieve evaluations in milliseconds. The number of facts by itself has very little impact on performance, unless you are inserting 1000s on each request. If you are able to design your application to insert facts up front, then a request may well be an insertion of a single fact. All benchmarks available achieve nothing more than showing whether a particular rules engine is a little bit faster than another at running a particular set of rules against a particular set of facts. And most are deliberately designed to be doing inefficient things, which are unlikely to bear any relation to real-world applications. So you need to run your own benchmarks against your own rules and facts. Steve On 24 Jul 2014, at 09:56, Zahid Ahmed <zahid.ahmed(a)emirates.com&gt; wrote: