One copy of the human genome contains about 3.1 billion base pairs. Because each base pair contains two nucleotides, that is about 6.2 billion nucleotides across the two DNA strands.
A typical human cell is diploid: it carries two genome copies, one from each parent. Before DNA replication, that means roughly 6.3 billion base pairs, or 12.6 billion nucleotides.
How many base pairs are in the human genome?
One haploid copy of the human genome contains about 3.1 billion base pairs. The GRCh38.p14 reference assembly totals 3,099,734,149 bp across all scaffolds and 3,088,269,832 bp across placed scaffolds.[1][2]
The count describes one set of 23 chromosomes, called a haploid genome. Sperm and egg cells are haploid. Most other nucleated human cells carry two chromosome sets and are diploid.
The Telomere-to-Telomere CHM13 assembly gives a related but different total. Its 2022 paper reported 3.055 billion base pairs for complete versions of chromosomes 1-22 and X, closing many repetitive regions left unresolved in earlier human genome sequencing references.[3]
For a general answer, 3.1 billion base pairs per haploid copy is the clearest figure. Technical comparisons should name the assembly and state whether the count includes all scaffolds, only placed chromosomes, gaps, or non-gap bases.
How many nucleotides are in human DNA?
A haploid human genome contains about 6.2 billion nucleotides across both DNA strands. A typical diploid cell contains about 12.6 billion nucleotides before it copies its DNA for cell division.
One base pair is a rung of the DNA double helix. It contains two nucleotides, one on each strand. Four types of nucleotide occur in DNA: adenine, cytosine, guanine, and thymine. The bases pair as A with T and C with G.[1]
| Genome count | Base pairs | Nucleotides across both strands |
|---|---|---|
| Haploid, one chromosome set | About 3.1 billion | About 6.2 billion |
| Diploid, two chromosome sets | About 6.3 billion | About 12.6 billion |
The nucleotide totals are calculated by multiplying the base-pair counts by two. Piovesan and colleagues calculated 6,270,605,410 bp for a reference 46,XY cell and 6,369,418,890 bp for a reference 46,XX cell. The difference comes mainly from the X chromosome being larger than the Y chromosome.[7]
These figures describe nuclear DNA before replication. Mature red blood cells have no nucleus, while sperm and egg cells are haploid. Mitochondrial DNA adds a much smaller, cell-dependent amount. The physical length of the same DNA is a separate measurement.
How large is the human genome as data?
The GRCh38.p14 human genome is about 3,099,734 kb, 3,100 Mb, or 3.10 Gb in genomics units. Stored as one plain-text character per base, the sequence takes about 3.10 GB, or 2.89 GiB, before FASTA headers and line breaks.
| Unit | Human genome size | What the unit means |
|---|---|---|
| bp | 3,099,734,149 bp | bp in GRCh38.p14 across all scaffolds |
| kb | 3,099,734 kb | Kilobases; 1 kb = 1,000 bp |
| Mb | 3,100 Mb | Megabases; 1 Mb = 1,000,000 bp |
| Gb | 3.10 Gb | Gigabases; 1 Gb = 1,000,000,000 bp |
| GB | About 3.10 GB | Gigabytes if one base is stored as one byte |
The bp, kb, Mb, and Gb rows are direct conversions from the Genome Reference Consortium's GRCh38.p14 all-scaffold length. The GB row is a storage calculation: one ASCII DNA letter takes one byte, so 3.099 billion letters take about 3.099 billion bytes before file-format overhead.[2]
A lowercase b counts biological bases; an uppercase B counts bytes. The human genome is about 3.1 Gb as sequence and about 3.1 GB as uncompressed one-letter-per-base text.
A compressed reference genome is usually under 1 GB. UCSC lists hg38.fa.gz at 938 MB and the compact hg38.2bit file at 797 MB.[4]
| File or data type | Typical size | What it contains |
|---|---|---|
hg38.2bit | 797 MB | Compact two-bit encoded hg38 reference |
hg38.fa.gz | 938 MB | Gzip-compressed hg38 FASTA reference |
| Plain sequence text | About 3.1 GB | One character per base, before headers and line breaks |
| Two-bit minimum | About 0.78 GB | Four DNA letters can be represented with two bits per base |
FASTA is a text format with a definition line for each sequence and lines of nucleotide letters. NCBI recommends nucleotide sequence lines no longer than 80 characters, so headers and line breaks add a small amount beyond the letters themselves.[5]
Raw whole-genome sequencing files are much larger. A 30x run reads the genome repeatedly; Illumina's NovaSeq X specifications assume more than 120 Gb of sequence data per sample. FASTQ files also store read names and quality scores.[6]
Why do different sources give different human genome sizes?
Human genome size estimates differ because they count different objects: one or two chromosome sets, a specific reference assembly, placed chromosomes or all scaffolds, gap characters or known bases, and biological sequence or computer files.
GRCh38.p14 illustrates the difference. Its all-scaffold total length is 3,099,734,149 bp, but its all-scaffold ungapped length is 2,948,611,470 bp because the ungapped count excludes N bases used for unresolved sequence. The placed-scaffold ungapped length is smaller again, at 2,937,639,396 bp.[2]
The rounded and exact answers can both be correct. “About 3.1 billion base pairs” describes a haploid human genome. “3,099,734,149 bp across all GRCh38.p14 scaffolds” identifies both the assembly and its counting rule.
Genome size is also separate from physical DNA length and gene count. The same 3.1-billion-base reference can be discussed as a file, a molecule, or an annotation system.
Sources▼
- Base Pair National Human Genome Research Institute · July 1, 2026. https://www.genome.gov/genetics-glossary/Base-Pair
- Human Genome Assembly GRCh38.p14 Genome Reference Consortium · July 1, 2026. https://www.ncbi.nlm.nih.gov/grc/human/data
- The complete sequence of a human genome Science · 2022. https://pubmed.ncbi.nlm.nih.gov/35357919/
- Index of /goldenPath/hg38/bigZips UCSC Genome Browser · July 1, 2026. https://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/
- FASTA Format for Nucleotide Sequences NCBI GenBank · July 1, 2026. https://www.ncbi.nlm.nih.gov/genbank/fastaformat
- NovaSeq X Specifications Illumina · July 1, 2026. https://www.illumina.com/systems/sequencing-platforms/novaseq-x-plus/specifications.html
- On the length, weight and GC content of the human genome BMC Research Notes · 2019. https://link.springer.com/article/10.1186/s13104-019-4137-z
Founder and computational chemist, ProteinIQ
Dr. Matic Broz is the founder of ProteinIQ and a computational chemist. He completed a PhD focused on protein structure, molecular dynamics, and neural networks, and writes about structural biology and scientific software.