The definition of an electroencephalographic (EEG)-based brain dysmaturity index that could allow automatic detection of neonates who deviate from expected ontogenetic patterns is proposed. The investigation was performed in a group of 94 term and preterm infants (28-112 weeks postconceptional age). For each neonate, one continuous two-channel EEG of 1-6 hours was recorded. The cluster analysis of different age groups was performed with a self-referential neural network. The network performed a nonlinear discriminant analysis; the synaptic strength of input nodes indicates the relevance of an individual EEG feature. The most relevant EEG features are given by the average amplitude in the delta and theta bands and by the relative amplitudes of beta-1/theta and beta-1/delta, respectively. The correlation between the frequency shifts and the postconceptional age agreed with measures of brain dysmaturity in healthy preterm neonates. Thus the presented trend in early EEG development demonstrates that it is possible to establish clinically relevant age dysmaturity scores.