Background An automated technique for recording eating and rumination behaviour was evaluated in ten lactating Brown Swiss cows by comparing data obtained from a pressure sensor with data obtained via direct observation over a 24-hour period. The recording device involved a pressure sensor integrated in the noseband of a halter. The analysed variables included number and duration of individual rumination, eating and resting phases, total daily length of these phases and number of cuds chewed per day. Results Eating and rumination phases were readily differentiated based on characteristic pressure profiles. Chewing movements during rumination were regular and generated regular waveforms with uniform amplitudes, whereas eating generated irregular waveforms with variable amplitudes. There was complete or almost complete agreement and no significant differences between data obtained via direct observation and pressure sensor technique. Both methods yielded an average of 16 daily eating phases with a mean duration of 28.3 minutes. Total time spent eating was 445.0 minutes for direct observation and 445.4 minutes for the pressure sensor technique. Both techniques recorded an average of 13.3 rumination phases with a mean duration of 30.3 (direct observation) and of 30.2 (pressure sensor) minutes. Total time spent ruminating per day, number of cuds per day and chewing cycles per cud were 389.3 and 388.3 minutes, 410.1 and 410.0 and 60.0 and 60.3 for direct observation and pressure sensor technique, respectively. There was a significant difference between the two methods with respect to mean number of chewing cycles per day (24′669, direct observation vs. 24′751, pressure sensor, P < 0.05, paired t-test). There were strong correlations between the two recording methods with correlation coefficients ranging from 0.98 to 1.00. Conclusions The results confirmed that measurements of eating and rumination variables obtained via the pressure sensor technique are in excellent agreement with data obtained via direct observation.