How to Read BLAST Results: E-value, Identity and Query Cover
9 min read · Updated September 24, 2026
A BLAST report gives every hit four numbers — bit score, E-value, percent identity and query cover — and each one answers a different question. Reading any one of them on its own is how people conclude their clone is a virus, their chimeric insert is a perfect match, or their primer has no match anywhere.
This guide explains the four numbers and then runs three real searches against GenBank, made on 24 September 2026 with SeqBench's BLAST tool, that each break one of them: a 600-base piece of EGFP, the same piece with 300 unrelated bases attached, and a single 21-base primer.
The four numbers in every hit
Identity and cover are read together, and the E-value tells you whether to take either seriously. A hit that is 100% identical over 100% of the query with a tiny E-value is a match; any weaker combination is a question.
- Bit score — the alignment's score, normalised so it can be compared between searches. Higher is a better alignment, and it does not depend on the size of the database.
- E-value — the number of hits at least this good you would expect to find by chance in a database of this size. It grows with the database and with the query length, so the same alignment gets a larger E-value in a bigger database. It is a measure of surprise, not of similarity.
- Percent identity — identical positions divided by the length of the alignment. It is computed over the aligned region only, and says nothing about the parts of your query that did not align.
- Query cover — the percentage of your query covered by the hit's aligned segments. It is the number that tells you whether a hit explains your whole sequence or only part of it.
Search 1: a clean match, and why the top hit is a virus
The first 600 bases of the EGFP coding sequence, searched with megablast against the default nucleotide database, return hits at 100% identity, 100% query cover, a bit score of 1,109 and an E-value reported as 0. That E-value is not literally zero; it is too small for the report to represent, which is normal for a perfect match several hundred bases long.
The top hit is not a GFP gene. It is the complete genome of an engineered African swine fever virus isolate whose genome contains EGFP, followed by cloning vectors, lentiviral constructs and 'synthetic construct' entries. The organism column names whatever GenBank record happens to contain your sequence, and a reporter like EGFP sits in thousands of them. The result says the query is EGFP; it says nothing about the virus.
Two more things in that list are easy to misread. Every one of the 20 hits scored identically, so their order is arbitrary rather than a ranking. And 20 is the number of hits the search was asked to return, not the number of matching records in GenBank, which for EGFP runs into the thousands.
Search 2: 100% identity is not a match
Take the first 300 bases of EGFP and join 300 random bases to the end. The top hits are still at 100% identity, and the E-value is still vanishingly small at 4.8 × 10⁻¹⁵³ — but query cover is 50%, and the bit score has halved to 555.
Identity is computed over the aligned stretch only, and every aligned base is EGFP. The other 300 bases matched nothing in the database. Read the identity column alone and you would call this sequence EGFP; the cover column says half of it is something else. In a real clone that half might be an insert, a vector junction, a contaminant or a chimera, and it is exactly the half you need to know about.
Search 3: a perfect primer match that is not significant
Search the 21-base EGFP forward primer ATGGTGAGCAAGGGCGAGGAG with megablast and the report is empty. Megablast starts each alignment from an exact seed of 28 bases, so a query shorter than 28 bases cannot seed anything at all. The blastn algorithm, with an 11-base seed, finds matches straight away.
Those matches are at 100% identity and 100% query cover — a perfect match — with an E-value of 5.3 and a bit score of 39. An E-value above 1 means chance alone would be expected to produce a hit that good in a database this size. That is not a contradiction: 21 bases is too short to be statistically surprising in a database the size of GenBank's, even though it is exactly long enough to prime a PCR.
The hit list says something useful anyway. At least three of the twenty hits are mCherry constructs that contain no GFP sequence at all, and the reason is in the sequence: the mCherry coding sequence (GenBank AY678264) begins with the same 21 bases as EGFP and ends with the same 24 bases the EGFP reverse primer anneals to. A primer pair designed for EGFP amplifies mCherry too, giving a 711 bp product beside EGFP's 720 bp — indistinguishable on an agarose gel. For a primer, BLAST answers where it can bind, not what your template is; for a proper primer-specificity check use a tool built for it, such as NCBI's Primer-BLAST.
What counts as a good E-value?
There is no universal cut-off, because the E-value depends on the size of the database and the length of the query. As a working guide: for identifying a clone or a PCR product you want near-100% identity and near-100% query cover, and E-values for full-length hits of a few hundred bases are then far below 10⁻⁵⁰. For homology searches between species, where identity is lower, thresholds such as 10⁻⁵ are common, but they are conventions rather than properties of the data.
SeqBench's BLAST tool, like the BLAST+ programs, reports hits up to an E-value of 10 by default. That is a reporting limit, not a significance threshold: anything with an E-value near 1 or above is what chance produces, and should be read that way.
When BLAST finds nothing
- The query is too short for megablast. Anything under 28 bases cannot seed a megablast search; use blastn, or a short-query mode, for primers and probes.
- The wrong program for the molecule. A protein query needs blastp or tblastn, and a DNA query that encodes a protein from a distant species is often found by blastx when blastn finds nothing.
- The sequence is genuinely diverged. Megablast is built to find near-identical sequences; a slower blastn search finds more distant ones.
- The database does not contain it. Since August 2024 NCBI's default nucleotide database has been core_nt, which leaves out most eukaryotic chromosome sequences; a hit that only exists in a genome assembly needs a genome database instead.
- Low-complexity masking. Simple repeats and poly-A stretches are masked before searching, so a query made mostly of them has little left to align.
BLASTing a Sanger read
The ends of a Sanger read are its lowest-quality bases, and miscalled bases there cost identity without changing what the sequence is. Trim the read to its high-quality region before searching, or read identity and cover with the trace's quality in mind: a 97% identity confined to the last 80 bases of a read is usually the chromatogram, not the clone.
Frequently asked questions
What does the E-value mean in BLAST?
The number of hits with a score at least as good as this one that you would expect to see by chance, given the size of the database and the length of your query. Smaller is more significant. A perfect 600-base match gets an E-value reported as 0; a perfect 21-base primer match got 5.3 in the same database, because matches that short occur by chance.
What is a good E-value for a BLAST hit?
It depends on what you are asking. To identify a clone or PCR product, look for near-100% identity and query cover, which for a few hundred bases comes with E-values far below 10⁻⁵⁰. For cross-species homology, cut-offs such as 10⁻⁵ are common conventions. An E-value near 1 or higher is what chance produces.
What is the difference between query cover and percent identity?
Percent identity is measured only over the part of your query that aligned; query cover is how much of your query aligned at all. A 600-base query whose first 300 bases are EGFP hits EGFP at 100% identity but only 50% query cover — the identity column alone would call it a perfect match.
Why is my top BLAST hit a virus or a vector when I sequenced a gene?
Because the organism column names whichever GenBank record contains your sequence, and widely used genes and markers sit inside thousands of vectors and engineered genomes. A 600-base piece of EGFP returns an engineered African swine fever virus genome as its top hit. Judge the hit by identity and cover, and read the organism as a description of that record, not of your sample.
Why does BLAST find nothing for my primer?
Because megablast, the default for nucleotide searches, needs an exact 28-base seed, and a primer is shorter than that. Run blastn instead, and expect a perfect match to come with a large E-value: a 21-base primer matched at 100% with an E-value of 5.3. For primer specificity, use a tool designed for it, such as Primer-BLAST.
What does the bit score mean?
It is the alignment's raw score rescaled so that it is independent of the scoring system and the database size, which makes it comparable between searches. Higher is better. In the searches in this guide, a 600-base perfect match scored 1,109 bits and a 300-base perfect match 555.
Related references
Related tools
Paste a sequence and find out what it is, against the whole of GenBank, with the program chosen for you.
Paste any unknown sequence and get a full identity report, then send it to BLAST.
Inspect chromatograms, compare a read with its reference, or screen a whole plate of clones.
Align two sequences with Needleman-Wunsch or Smith-Waterman.
Paste a GenBank, RefSeq or UniProt accession and get the FASTA or GenBank record.