A comprehensive model on heat transfer in welded plates is able to calculate the amount of heat losses from the surfaces. A model demands as input parameter the amount of heat delivered to the plate, independently of any loss (called here gross heat input for clarity). However, the great discrepancies among the results of calorimetric measurements have left many researchers skeptical about using this parameter in modeling as absolute term. The objective of this work was to assess the use of a water-cooled stationary anode calorimeter to obtain not only arc efficiency, but also gross heat input. A series of tests was carried out to determine the effect of current, material type and water flow rate on the calorimeter performance, as well as to evaluate some measures for reducing the calorimeter intrinsic errors. Finally, a sensitivity test was conducted to estimate the effect of measurement inaccuracies on the absorbed heat and arc efficiency values. The results showed that this calorimetric approach is a simple way for measuring gross heat inputs in arc welding. Nevertheless some improvement to reduce heat losses from the top surface and boost heat sinking from the opposite surface of the test coupon must be applied. This calorimeter is, on the other hand, highly sensitive to the parameter measurements, leading to errors up to ± 0.09 in arc efficiency determination if the instrument is not properly calibrated and installed.