Despite its benefits, grazing is decreasing , in southern Belgium because of difficult management. In fact, farmers lack confidence in the amount of fodder available on the grasslands. The herbage mass (HM - kg DM/ha) is an essential data to estimate grass stocks. To obtain it, the reference method is to clip grass from delimited quadrats in the paddocks. The weight of the sample (kg DM/quadrat) allows an estimate of the stocks (kg DM/ha) on the paddock. However, this method needs labour to get a valuable overview of the grassland stocks. Another method is based on measuring the sward height (H-cm) using rising plate meters (RPM) and then to convert it into HM. Using the RPM is easy but needs a calibration equation to convert the compressed height (CSH, cm) to HM. Yet, the rising plate meters available on the market are calibrated by manufacturers under conditions somewhat different from those where they will be used. This study aims thus to evaluate the accuracy of different calibration equations from commercial RPM. Measurements of CSH and HM were carried out in 5 Walloon farms on permanent grasslands from 2013 to 2015. In total, 299 data were collected. We compared three calibration equations. Equation 1 developed in New-Zealand is proposed at use of Jenquip EC20®. Equation 2 developed in Ireland is proposed at use of GrassHopper® and the Equation 3 is developed in France. Calculated HM was faced to the collected field data. The coefficient of determination (R²), the mean square error (MSE) and the relative prediction error (RPE) were determined. The HM and CSH of field data were 1599 ± 674 kg DM/ha and 9.73 ± 2.86 cm (mean ±SD). The values of HM obtained from Equ.1 to 3 were: 1862 ± 400 kg DM/ha (Equ. 1), 1489 ± 339 kg DM/ha (Equ. 2) and 2315 ± 680 kg DM/ha (Equ. 3). The R² values ranged from 0.41 (Equ.2) to 0.42 for Equ.1-3. The Equ.1 and 3 led to a recurrent over-evaluation of stocks (74% in Equ.1 and 90% in Equ. 3). This over-estimation reached 503 kg DM/ha. Conversely, the RPE of Equ. 2 was the lowest (33.6%) and the errors were more equally balanced (55% under vs 45% over-estimation). It seems thus that the Irish equation could be fairly used.