Clinically significant copy-number variants (CNVs) known to cause human diseases are routinely detected by chromosomal microarray analysis (CMA). Recently, genome sequencing (GS) has been introduced for CNV analysis; however, sequencing depth (determined by sequencing read-length and read-amount) is a variable parameter across different laboratories. Variating sequencing depths affect the CNV detection resolution and also make it difficult for cross-laboratory referencing or comparison. In this study, by using data from 50 samples with high read-depth GS (30×) and the reported clinically significant CNVs, we first demonstrated the optimal read-amount and the most cost-effective read-length for CNV analysis to be 15 million reads and single-end 50 bp (equivalent to a read-depth of 0.25-fold), respectively. In addition, we showed that CNVs at mosaic levels as low as 30% are readily detected, furthermore, CNVs larger than 2.5 Mb are also detectable at mosaic levels as low as 20%. Herein, by conducting a retrospective back-to-back comparison study of low-pass GS versus routine CMA for 532 prenatal, miscarriage, and postnatal cases, the overall diagnostic yield was 22.4% (119/532) for CMA and 23.1% (123/532) for low-pass GS. Thus, the overall relative improvement of the diagnostic yield by low-pass GS versus CMA was ~ 3.4% (4/119). Identification of cryptic and clinically significant CNVs among prenatal, miscarriage, and postnatal cases demonstrated that CNV detection at higher resolutions is warranted for clinical diagnosis regardless of referral indications. Overall, our study supports low-pass GS as the first-tier genetic test for molecular cytogenetic testing.