Children reading fluency has gained focus on the last few years with the increasing importance that governments give to this subject, as shown by the implementation of curricular goals (CG) for the Portuguese case. Also, the issue of automatic evaluation of children’s reading performance by detecting and analyzing errors and disfluencies in speech is important to build automatic reading tutors. It can also aid and complement the current manual and subjective method for evaluations of overall reading ability in schools. In this paper we describe the acquisition procedure and analysis of a read-aloud speech database of European Portuguese from children aged 6-10 from the first to the fourth school grades that allows the evaluation of reading performance. We also intend to fill a gap in the availability of Portuguese databases of children reading aloud. The collected amount of speech exhibits a large variety of errors and disfluencies, which assumes an important role in the tasks of training acoustic, disfluency and pronunciation models for automatic reading assessment systems. Results of average reading performance in the database are approximately in line with the government defined curricular goals for reading in the 1st four years of the primary school.