Heerema Marine Contractors (HMC) owns and operates several semi-submersible crane vessels (SSCVs), used for offshore heavy lift operations. In preparation of these operations, hydrodynamics analyses in the frequency domain are conducted using software developed in-house at HMC. These analyses result in Response Amplitude Operators (RAOs), which are used for estimating dynamic response to a certain sea state. Crane hook load fluctuations response modelled in this manner is generally larger than measured, however. Since hook load is often a limiting criterion, an overestimation of hook load fluctuations in the models could undesirably affect perceived operability. Three possible root causes for this discrepancy have been identified: 1. Modelled hydrodynamics are inaccurate 2. Measured dynamic tension does not represent dynamic hook load 3. Dynamic tension measuring equipment is inaccurate In order to assess root cause (1.), a comparison study is conducted. Hydrodynamic rigid body models are created for four stages of a typical heavy lift operation: the free-floating, lift-off, free-hanging and set-down stages. Wave forcing, added mass and damping properties for the floating bodies are obtained from diffraction analyses performed using WAMIT. Model sensitivity to different hydrodynamic properties, obtained from single- vs. multi-body and infinite vs. actual water depth diffraction analyses, is tested as well. Measurements have been obtained for a project where a topsides module was lifted from a barge onto a jacket support structure. Measuring equipment included 6DoF motion sensors on the SSCV, module and in one of the SSCV's crane boom tips. Generally, modelled motion response is found to be quite accurate. Originally, discrepancies between modelled and measured hook load fluctuations spectra were observed during the lift-off stage of a heavy lift operation. However, measured hook load fluctuations prove to be smaller than modelled for all stages considered. Hook load fluctuations are governed by relative vertical motions of the module and the SSCV's crane boom tip. Magnitude and phasing of these signals imply over- and underestimations of wire rope elasticity and damping properties respectively. Furthermore, the ratio between measured hook load fluctuations and hook load fluctuations corresponding to measured relative motions is quite different per stage. A stage dependent contribution of root causes (2.) and/or (3.) is therefore expected to add to the overall discrepancy. In an attempt to obtain accurate crane wire rope elasticity properties from a simplified 2DoF dynamic crane model, natural frequencies during the free-hanging stage are searched for. Implementing these frequencies in the 2DoF system allows for two unknowns to be solved: a stiffness for the boom suspension wires, and an equivalent stiffness for the grommets and main hoist wires. Response peaks found within a frequency range corresponding to probable stiffnesses of the wire rope sections in the system are not very explicit. For the most probable and explicit response peaks found, no combination of stiffnesses exists for which this 2DoF model is solvable. These response peaks therefore cannot be explained by the modes searched for, and stiffness properties cannot be obtained from this model unfortunately. Therefore, elasticity properties for wire rope elements in the model are obtained from a stiffness test in literature. Subsequently, lift object vertical motions time traces are modelled, using imposed (measured) motions of the SSCV's crane tip as input. Alternative damping values are applied in order to minimize phasing between modelled and measured vertical motions of the lift object. This yields a supercritical damping value, whereas damping values are set at 1.5\% of the critical damping in the models by default. Applying the new elasticity and damping properties in the models results in a significant change in modelled hook load fluctuations. However, correspondence to measured hook load fluctuations does not necessarily increase, and discrepancies remain. Furthermore, the measured signals used contain quite some noise, and root causes (2.) and/or (3.) are expected to add to the overall discrepancy observed. An assessment of root causes (2.) and (3.), and repetitions of this research for other lifts using more accurate measuring equipment are therefore recommended.