Background: The genetic basis of left ventricular (LV) image-derived phenotypes, which play a vital role in the diagnosis, management and risk stratification of cardiovascular diseases, is unclear at present. Methods: The LV parameters were measured from the cardiovascular magnetic resonance (CMR) studies of the UK Biobank. Genotyping was done using Affymetrix arrays, augmented by imputation. We performed genome-wide association studies (GWASs) of six LV traits - LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LV stroke volume (LVSV), LV ejection fraction (LVEF), LV mass (LVM) and LV mass to end-diastolic volume ratio (LVMVR). The replication analysis was performed in Multi-Ethnic Study of Atherosclerosis (MESA). We identified the candidate genes at GWAS loci based on the evidence from extensive bioinformatic analyses. Polygenic risk scores (PRSs) were constructed from the GWAS summary statistics to predict the heart failure events. Results: The study comprised 16,923 European UK Biobank participants (mean age: 62.5 years; 45.8% men) without prevalent myocardial infarction or heart failure. We discovered fourteen genome-wide significant loci - three loci each for LVEDV, LVESV and LVMVR; four loci for LVEF and one locus for LVM - at a stringent p < 1 x 10-8. Three loci were replicated at Bonferroni significance and seven loci at nominal significance (P < 0.05 with concordant direction of effect) in the MESA study (N = 4,383). Follow-up bioinformatic analyses identified 28 candidate genes which were enriched in the cardiac developmental pathways and regulation of the LV contractile mechanism. Eight genes (TTN, BAG3, GRK5, HSPB7, MTSS1, ALPK3, NMB and MMP11) supported by at least two independent lines of in-silico evidence were implicated in the cardiac morphogenesis and heart failure development. The PRSs of LV phenotypes were predictive of heart failure in a hold-out UK Biobank sample of 3,106 cases and 224,134 controls (odds ratio 1.41, 95% CI: 1.26 - 1.58, for the top quintile vs the bottom quintile of the LVESV risk score). Conclusions: We report fourteen genetic loci and indicate several candidate genes which not only enhance our understanding of the genetic architecture of prognostically important LV phenotypes but also shed light on potential novel therapeutic targets for LV remodelling.