Optical properties of GaAsN system with nitrogen concentrations in the range of 0.9-3.7% are studied by full-potential LAPW method in a supercell approach. The E+ transition is identified by calculating the imaginary part of the dielectric function. The evolution of the energy of this transition with nitrogen concentration is studied and the origin of this transition is identified by analyzing the contributions to the dielectric function from different band combinations. The L1c-derived states are shown to play an important role in the formation of the E+ transition, which was also suggested by recent experiments. At the same time the nitrogen-induced modification of the first conduction band of the host compound are also found to contribute significantly to the E+ transition. Further, the study of several model supercells demonstrated the significant influence of the nitrogen potential on the optical properties of the GaAsN system. GaAs 1−x N x alloys have attracted much attention of both experimentalists and theoreticians due to their unusual physical properties. Among those features, observed in the typical nitrogen concentration regime of 0.5%-3.0%, one should mention the anomalous band gap reduction, which makes this system technologically important for such electronic devices as infrared diode lasers [1] and multijunction solar cells [2]. Another striking feature of this system is the appearance of a new composition-dependent optical transition, called E + , detected by electro-reflectance measurements for the samples with nitrogen concentrations of ≥0.8% [3,4]. This peak is located at about 0.4-0.8 eV above the conduction band (CB) minimum (denoted as E − ) and shifts upward in energy with increase of nitrogen concentration [3,4,5].Several physical mechanisms have been proposed to describe these nitrogen-induced changes : two-level band anticrossing (BAC) model [6,7], disorder-allowed Γ-X-L coupling [8,9], and the formation of the impurity band [10]. In the frames of the BAC model the E + state naturally appears as the high-energy solution of a two-state Hamiltonian, describing the interaction of the localized nitrogen state with the extended CB states of the host matrix compound [11]. However, the firstprinciples [8,12,13,14] as well as empirical pseudopotential [9,15,16] calculations failed to validate the BAC model. On the other hand, based on these calculations, the intraband coupling model has been proposed. In this model the E + peak is caused by a disorder-allowed transition from the valence band (VB) maximum to the Lpoint of the conduction band. In particular, it was proposed that the energy position of the E + transition is a configuration-weighted average of a nitrogen impurity state and CB L-state, denoted as L 1c [9]. However, to our knowledge, there are no studies of the GaAsN system, which identify and clarify the origin of the E + transition based on direct first-principles calculations of the optical properties of this system.In this Letter we report the results of theoretical study of the...