Using an internet collected dataset, we will provide some empirical evidence on the information that Dutch high school students possess before their decision on tertiary education participation. The sample is prone to selective participation and high attrition, but we detect little systematic effects and inconsistent reporting of probability distribution is not more frequent than in controlled settings. We find little support for patterns that one would expect from individuals having private information. Girls expect substantially lower earnings in all schooling scenarios, but implicit rate of returns do not differ from those anticipated by boys. We find no evidence of expected risk compensation.