The objective of this work was to elucidate whether a sperm selection method that combines rheotaxis and microfluidics can improve the selection of spermatozoa over density gradient and swim-up. For this purpose human sperm selected by rheotaxis were compared against density gradient, swim-up and a control group of non-selected spermatozoa in split frozen-thawed (FT) and fresh (F) semen samples. Sperm quality was assessed in terms of motility, morphology, DNA fragmentation index (DFI), viability, acrosome integrity and membrane fluidity. Using a mouse model, we compared fertilisation and embryo development rates after performing ICSI with spermatozoa, sorted using rheotaxis or swim-up. Selection by rheotaxis yielded a sperm population with reduced DFI than the control (P < 0.05), improved normal morphology (P < 0.001) and higher total motility (TM; P < 0.001) than the other techniques studied in F and FT samples. Swim-up increased TM compared to density gradient and control in FT or F samples (P < 0.001), and yielded lower DFI than the control with F samples (P < 0.05). In FT samples, selection by rheotaxis yielded sperm with higher viability than control, density gradient and swim-up (P < 0.01) while acrosomal integrity and membrane fluidity were maintained. When mouse spermatozoa were selected for ICSI using rheotaxis compared to swim-up, there was an increase in fertilisation (P < 0.01), implantation (P < 0.001) and foetal development rates (P < 0.05). These results suggest that, in the absence of non-destructive DNA testing, the positive rheotaxis can be used to select a population of low DNA fragmentation spermatozoa with high motility, morphology and viability, leading to improved embryo developmental rates.