Background: Global longitudinal strain (GLS) is a predictor of outcome after cardiac surgery. If integrated into clinical decision-making and timing of surgery, it is important to evaluate the feasibility, reproducibility, and variation of GLS in this selection of patients, where poor image quality and nontraceable segments are frequent. Methods and results: Two-dimensional strain analysis was performed on 250 patients planned to undergo open-heart surgery. Intra- and inter-examiner retest variability was assessed in 50 consecutive patients. All myocardial segments were traceable in 119 patients, and GLS of those served as a reference in comparison with alternative strain models with nontraceable segments. Global longitudinal strain estimation by the recommended method of a maximum of one nontraceable segment per view was only feasible in 64% of cases (mean GLS −16%). Reproducibility was moderate (intra-observer coefficient of variation [CV] 8%; inter-observer CV 10%) and variation of GLS showed bias ± 95% limits of agreement (LOA) of 0.6 ± 1.1 (P <.05). Accepting three nontraceable segments in total increased feasibility to 77% with similar reproducibility (intra-observer CV 8%; inter-observer CV 11%) and variation (bias ± LOA: 0.6 ± 1.3, P <.05). A model with a maximum of one apical, one mid, and one basal nontraceable segment increased feasibility to 72% with similar reproducibility (intra-observer CV 8%; inter-observer CV 10%) and variation (bias ± LOA: 0.4 ± 1.2, P <.05). Conclusion: Global longitudinal strain estimation in patients prior to cardiac surgery is challenged by moderate feasibility, retest variation as well as variation in cases of nontraceable segments. We suggest alternative strain models with improved feasibility without compromising reproducibility and variation.