A large number of in vivomeasurements of forage digestibility and intake is difficult to obtain, especially for permanent grasslands that cannot easily be studied outside their location. A study was conducted to estimate the variability of forage in vivo digestibility and voluntary intake in sheep between and within experimental stations. Bales from two permanent grassland hays harvested on the same plots were delivered to five experimental stations. Samples were analysed for ash, crude protein and neutral detergent fibre. Digestibility and intake trials for both hays were performed simultaneously in the five experimental stations. Repeatability (error within laboratories) and reproducibility (error between laboratories) for these measures were assessed. Chemical composition variability between forage samples at each station was important for ash. For dry matter digestibility (DMD), organic matter digestibility (OMD) and neutral detergent fibre digestibility (NDFD), the variability between animals was greater than between stations and repeatability and reproducibility values were close (0.04 and 0.05 respectively for OMD). For dry matter intake (DMI) and digestible organic matter intake (DOMI) the variability of the measurement was more than three times greater within an experimental station and more than four times greater between experimental stations than for digestibility. For DMI the variability associated with animals within a station was also the greatest source of variation in reproducibility. However, for this measure, the between-stations variability was greater than that found for digestibility. The repeatability and reproducibility obtained for the DOMI were similar to those obtained for the DMI (15.20 and 20.45 respectively for DMI and 10.81 and 13.76 for DOMI). Our results suggest that data obtained in different locations for the in vivo OMD using the same methodology, may be pooled without significantly decreasing the precision of the measurement. However, precision for the DMI might be adversely affected by pooling data obtained in different stations.