Context : The emission process responsible for the so–called “ prompt ” emission of gamma–ray bursts is still unknown .
A number of empirical models fitting the typical spectrum still lack a satisfactory interpretation .
A few GRB spectral catalogues derived from past and present experiments are known in the literature and allow to tackle the issue of spectral properties of gamma–ray bursts on a statistical ground .

Aims : We extracted and studied the time–integrated photon spectra of the 200 brightest GRBs observed with the Gamma–Ray Burst Monitor which flew aboard the BeppoSAX mission ( 1996–2002 ) to provide an independent statistical characterisation of GRB spectra .

Methods : The spectra have a time-resolution of 128 s and consist of 240 energy channels covering the 40–700 keV energy band .
The 200 brightest GRBs were selected from the complete catalogue of 1082 GRBs detected with the GRBM ( Frontera et al .
2009 ) , whose products are publicly available and can be browsed/retrieved using a dedicated web interface .
The spectra were fit with three models : a simple power–law , a cut–off power law or a Band model .
We derived the sample distributions of the best-fitting spectral parameters and investigated possible correlations between them .
For a few , typically very long GRBs , we also provide a loose ( 128-s ) time–resolved spectroscopic analysis .

Results : The typical photon spectrum of a bright GRB consists of a low–energy index around 1.0 and a peak energy of the \nu F _ { \nu } spectrum E _ { p } \simeq 240 keV in agreement with previous results on a sample of bright CGRO/BATSE bursts .
Spectra of \sim 35 % of GRBs can be fit with a power–law with a photon index around 2 , indicative of peak energies either close to or outside the GRBM energy boundaries .
We confirm the correlation between E _ { p } and fluence , in agreement with previous results , with a logarithmic dispersion of 0.13 around the power–law with index 0.21 \pm 0.06 .
This is shallower than its analogous in the GRB rest–frame , the Amati relation , between the intrinsic peak energy and the isotropic–equivalent released energy ( slope of \sim 0.5 ) .
The reason for this difference mainly lies in the instrumental selection effect connected with the finite energy range of the GRBM particularly at low energies .

Conclusions : We confirm the statistical properties of the low–energy and peak energy distributions found by other experiments .
These properties are not yet systematically explained in the current literature with the proposed emission processes .
The capability of measuring time–resolved spectra over a broadband energy range , ensuring precise measurements of parameters such as E _ { p } , will be of key importance for future experiments .