What Genetic Testing Costs, by Test Type

A genetic test costs somewhere between $0 and about $5,000, and the spread has almost nothing to do with quality. As of 2026, rough market prices run as follows. Consumer SNP array genotyping runs $99–$199, and a targeted clinical gene panel runs $250–$1,500. A clinical exome with interpretation runs $1,000–$2,500. A research-grade 30x whole genome with raw files runs $300–$1,000, and a clinical whole genome with a signed report from a CLIA/CAP lab runs $1,500–$3,000. Free testing exists and is real, mostly sponsor-funded programs for specific indications. The variable that moves price most is not how much DNA gets read, it is how much human interpretation and regulatory overhead is attached to the result.

That distinction matters for anyone who plans to do their own analysis. If you want files to work with, you are buying a commodity that has been falling in price for twenty years. If you want a clinician-signed interpretation, you are buying labor, and labor prices do not fall.

Where the money goes

The sequencing itself is cheap and getting cheaper. The $1,000 genome was a stated target from Illumina in the mid-2000s, back when a genome cost seven figures.¹ On a NovaSeq X with 25B flow cells, consumable cost for a 30x human genome is now in the low hundreds of dollars, and the marginal cost of adding one more sample to a full flow cell is lower still. That is why direct-to-consumer WGS at $400–$700 is a viable business and not a loss leader.

What you are paying for above that floor:

• Sample logistics: kit, shipping, accessioning, DNA extraction, and QC. $50–$150.
• Compute and storage: alignment plus variant calling is roughly 1–2 hours on a DRAGEN FPGA or 8–20 CPU-hours on commodity hardware. Under $50 at scale. Storage of a 30x CRAM is about 15–20 GB. On S3 Glacier Deep Archive that is under $0.30 a year.
• Interpretation: a board-certified geneticist or genetic counselor reviewing filtered variants against ACMG criteria, writing a report, and being liable for it. This is the line item that costs $500–$2,000 and it is the reason two labs running the same chemistry quote prices 5x apart.
• Regulatory and accreditation overhead: CLIA, CAP and sometimes NY State CLEP, plus orthogonal confirmation of reported variants by Sanger or a second assay.

Systematic review of health economic studies on exome and genome sequencing found reported costs varying widely across studies with inconsistent methods, and relatively few full economic evaluations that would let you compare like with like.² When you see a cost figure quoted in a paper or a press release, check whether it means reagents, reagents plus labor, or the whole diagnostic pathway including the clinic visits.

What each price tier gets you as files

This is the part most pricing pages skip. Ask before you pay.

SNP array, $99–$199. Illumina Global Screening Array or similar, roughly 650,000–750,000 probes. You get a tab-delimited text file of rsID, chromosome, position and genotype, around 15–25 MB. The array interrogates about 0.02% of your 3.1 billion bases and is blind to anything not on the chip, including most rare pathogenic variants. Imputation against a reference panel (TOPMed, HRC) extends this to tens of millions of common variants at usable accuracy, but imputed rare variants (MAF < 0.5%) are unreliable and should never be treated as a finding. Good for polygenic scores and ancestry. Poor for anything clinical.

Targeted panel, $250–$1,500. Hybrid capture or amplicon over 20–500 genes, typically 100–500x coverage. You should receive a report. Ask explicitly whether you can get the BAM and the VCF, since many clinical labs will not release them without a request, and some charge for it. Panels are the right instrument when there is a specific clinical question, because deep coverage plus curated gene content beats broad shallow data for sensitivity.

Exome, $1,000–$2,500 clinical, $400–$900 research-grade. Capture of roughly 1–2% of the genome at 100x mean. Real coverage is uneven: a meaningful fraction of coding exons fall below 20x on standard captures, GC-rich first exons worst of all. Ask for the coverage report. The mean alone is not enough.

Whole genome, 30x. The first file is FASTQ, about 100 GB gzipped for a paired-end 30x genome. The second is BAM or CRAM aligned to GRCh38 or T2T-CHM13. The third is a VCF or gVCF of about 4.5–5 million variants. Insist on all three tiers if you are doing your own work. FASTQ lets you realign to a new reference in five years without re-sequencing, which is the single most valuable property of the purchase.

We would buy the 30x genome with raw files, and would pay the premium for a clinical-grade genome only when there is a specific medical question and a clinician in the loop. The reason is simple: reanalysis is free and interpretation ages fast. Variant classifications change, gene-disease associations get added, and a five-year-old report is a five-year-old snapshot of a database. Your FASTQ is not.

If you plan to analyze it yourself

A workable stack, all open source:

bwa-mem2 mem -t 32 -R '@RG\tID:1\tSM:sample\tPL:ILLUMINA' GRCh38.fa r1.fq.gz r2.fq.gz \
  | samtools sort -@ 8 -o sample.bam -
samtools index sample.bam
# variant calling
run_deepvariant --model_type=WGS --ref=GRCh38.fa --reads=sample.bam \
  --output_vcf=sample.vcf.gz --num_shards=32
# annotate
vep -i sample.vcf.gz --cache --assembly GRCh38 --everything --vcf -o sample.vep.vcf
# sanity check depth
mosdepth -n --fast-mode -t 4 sample sample.bam

Budget about 20–40 CPU-hours for alignment and 4–8 for DeepVariant with a GPU, or overnight on a 32-core box. Check mosdepth output before anything else: mean depth below 28x, or more than 5% of the callable genome under 10x, means you got a thin run and should ask the lab about it.

Benchmark your pipeline once against a truth set. Download GIAB HG002, run your exact pipeline, and score with hap.py against the v4.2.1 high-confidence BED. You should see SNV F1 above 0.995 and indel F1 in the 0.99 range inside high-confidence regions. If you get materially worse, the problem is your pipeline. The biology is not the cause.

What price does not buy

Short-read WGS at 30x misses things regardless of what you pay. Repeat expansions (HTT, FMR1, C9orf72) need targeted PCR or long reads. Genes with near-identical pseudogenes (SMN1, PMS2, CYP21A2) are routinely miscalled by short reads and need MLPA or specialized callers. Structural variants and mid-size CNVs are detected poorly. Mitochondrial heteroplasmy below about 10% is unreliable. A lab quoting you “whole genome” is quoting you the 90% of the genome short reads handle well.

The other limit is interpretive. The cause is not technical. Sequencing healthy adults turns up variants of uncertain significance in large numbers, and the clinical value of genome sequencing in people without symptoms remains an open question rather than a settled one.³ European professional guidance has argued for targeted analysis of sequencing data rather than open-ended interrogation, precisely because breadth generates findings nobody can act on.⁴ A cheap file does not make a cheap answer. If a variant in your data suggests a medical risk, take it to a genetic counselor or medical geneticist before acting on it, and expect them to want orthogonal confirmation of the call.

Questions people also ask

Can I get free genetic testing? Yes, under specific conditions. Sponsored programs funded by pharmaceutical companies or research consortia cover testing for suspected rare disease, certain cancers, and cardiomyopathies at no cost to the patient. Large research cohorts (All of Us, UK Biobank participants, various academic biobanks) also return some results. The tradeoff is eligibility criteria you must meet and, in the sponsored case, data sharing with the sponsor. Read the consent document.

Will insurance pay for genetic testing? For a specific diagnostic indication with documented family or clinical history, often yes, subject to prior authorization and a deductible. For screening a healthy person out of curiosity, generally no. Exome and genome sequencing are covered in more places than five years ago, mostly for pediatric rare disease and oncology, but coverage policies remain inconsistent across payers.

Is genetic testing 100% accurate? No. In well-covered regions of the genome, modern short-read sequencing calls SNVs at above 99.5% sensitivity and precision against reference truth sets. Outside those regions, accuracy degrades sharply. Separately, a technically correct variant call can still be interpreted incorrectly, and roughly 1–3% of clinically reported variants get reclassified over time.

Why is one lab $199 and another $2,000 for what sounds like the same test? Depth of coverage, gene content, whether structural variants and CNVs are assessed, whether findings are orthogonally confirmed, and whether a licensed professional signs the report. Ask for the coverage specification and the confirmation policy. Those two answers explain most of the price gap.

Footnotes

1. Simon T. Bennett, C.L. Barnes, Anthony J. Cox, et al. Toward the $1000 Human Genome. Pharmacogenomics, 2005. https://doi.org/10.1517/14622416.6.4.373 ↩

2. Katharina Schwarze, James Buchanan, Jenny C. Taylor, et al. Are whole-exome and whole-genome sequencing approaches cost-effective? A systematic review of the literature. Genetics in Medicine, 2018. https://doi.org/10.1038/gim.2017.247 ↩

3. Noralane M. Lindor, Stephen N. Thibodeau, Wylie Burke. Whole-Genome Sequencing in Healthy People. Mayo Clinic Proceedings, 2017. https://doi.org/10.1016/j.mayocp.2016.10.019 ↩

4. Carla van El, Martina C. Cornel, Pascal Borry, et al. Whole-genome sequencing in health care. European Journal of Human Genetics, 2013. https://doi.org/10.1038/ejhg.2013.46 ↩