Bilateral fittings of bone-anchored hearing aids (BAHA) were evaluated in 25 patients with at least 3 months experience with using two BAHAs. For all patients, air conduction hearing aids were contraindicated due to either recurrent otorrhoea or otitis externa (19 cases) or to congenital aural atresia (six cases). Candidacy for bilateral fitting was primarily based on symmetry of bone conduction thresholds. For all patients, measurements comprised sound localisation, speech recognition in quiet and in noise. In addition, in a subgroup of nine patients, release from masking for pure-tone stimuli in noise with interaural phase differences (binaural masking level difference. BMLD) was measured. The percentage of correct localisation judgments with 500-Hz and 2-kHz noise bursts increased significantly (p<0.01) from 22.2 per cent and 24.3 per cent for unilateral fittings to 41.8 per cent and 45.3 per cent for bilateral fittings, respectively. With unilateral fittings sound localisation judgments appeared to be strongly biased to the ipsilateral BAHA side. whereas with bilateral fittings, judgments were far more symmetrical. The speech reception threshold for sentences in quiet was significantly (p<0.01) better for the bilateral fittings compared to the unilateral fittings: 37.5 dBA versus 41.7 dBA. Speech recognition in noise was measured with the speech signal presented in front of the listener and a 65-dBA masking noise at either +90 degrees or -90 degrees azimuth. For noise presented at the ipsilateral side of the first fitted BAHA, the signal-to-noise ratio was significantly reduced (p<0.01) from -0.7 dB for the unilateral fitting to -4.0 dB for the bilateral fitting. The speech reception threshold in noise was not significantly different (p>0.05) for unilateral and bilateral fittings when the noise was presented at the contralateral side of the first fitted BAHA. The results for the six patients with congenital atresia are comparable with those for the other patients. So, directional hearing and speech recognition in noise improve significantly with a second BAHA. The BMLD measurements showed a significant (p<0.01) release from masking of 6.1, 6.0 and 6.6 dB for 125-Hz, 250-Hz and 500-Hz stimuli, respectively. The BMLD effect of 4.1 dB at 1,000 Hz was not significant at the 5 per cent level. The positive results with the bilateral fittings in quiet can be ascribed to increased stimulus levels due to diotic summation of signals from either side. The results for localisation, speech recognition in noise and BMLD measurements indicate that bilaterally fitted BAHAs do indeed (to some extent) result in binaural hearing.