The fracture toughness of submerged arc weld metal on C-Mn-steel has been studied. The toughness was varied by adding different amounts of titanium, and the COD properties were measured at different positions in the weld metal and the HAZ. The welds were made in one layer from one side in a 30 deg y-groove without rootgap. The plate thickness was 30 ram and the groove depth 20 mm. The welding conditions were about 1200 A, 40 V, 30 cm/min travel speed. Some of the analyses and mechanical properties are summarized in Tables l and 2. The COD measurements were performed in accordance with [i]. Crosshead speed was 1 mm min -1In order to determine the onset of slow stable cracking, electrical potential measurements were performed.The potential over the crack was measured by means of two iron wires, that were spot welded on each side of the notch, Fig. i.The notch position in the welds was varied from the middle of the weld metal to the HAZ. The dependence for ¢i is evident, with low toughness in the middle of the weld metal, Fig. 2. The toughness increases just outside the middle zone, but decreases steadily to the fusion boundary and HAZ. The position dependence is most pronounced in weld A (0.004 percent Ti) with the highest ductility.The 6 i values are very close in the HAZ; this is what to expect since the bsse metal and welding conditions are the same for the welds. When 6 i becomes low, the measurements and calculations are more uncertain, and the values must not be taken as absolute. The 8max curves shows more disturbance, weld L (0.21 percent Ti) has another tendence than the other two, and the HAZ values varies.The typical weld metal primary structure is shown in Fig. 3. The structure is very orientated, and changes from fine cellular to coarse dendritic as the solidification advances.The best crack opening displacement parameter is a moderate extension of linear elastic fracture condition and should be related to fracture in absence of ductile crack growth. But to extend the application of elastic-plastic fracture mechanics, the allowed degree of crack tip deformation has to be increased. A useful parameter should therefore be the COD at initiation of brittle fracture after limited ductile crack growth.But mild-and low-alloyed steels are often tough, and the only COD parameters that can be measured are COD at initiation of ductile crack growth ~i" and COD at maximum load for steady ductile crack growth ~max-~i can be compared with yield strength in tension testing. The resistance against deformation is measured, but the onset of complete fracture is not predicted.Cmax is very easy to define and has often been used as a COD parameter.The degree of crack growth at maximum load is usually not known, and is used [i] in the same way as is ~i" The Int Journ of Fracture 15 (1979)