How to count matches with grep

Counting search results with grep is useful when a log check, report review, or command-output scan needs a number instead of matching text. The important choice is whether the number should represent matching lines or every occurrence of the pattern.

The -c option counts lines that contain at least one match. If the same pattern appears twice on one line, that line still adds only one to the count, so use grep -o with wc -l when the job is to count individual matches.

The sample below uses a short log where one line contains ERROR twice and one line contains lowercase error. That fixture makes the difference visible: grep -c 'ERROR' returns three matching lines, grep -o 'ERROR' | wc -l returns four uppercase occurrences, and grep -ci 'error' counts four lines after ignoring case.

Steps to count matches with grep:

Create a small sample log with repeated uppercase matches, a lowercase match, and a nonmatching line.

$ cat > app.log <<'EOF'
INFO service ready
ERROR api timeout
WARN retry queued
ERROR disk full ERROR retry failed
error lowercase diagnostic
ERROR api timeout
EOF

Count lines that contain the uppercase pattern.
```
$ grep -c 'ERROR' app.log
3
```
grep -c counts matching lines. The line ERROR disk full ERROR retry failed contains two matches, but it still contributes one line to this result.
Count every uppercase occurrence when repeated matches on the same line matter.
```
$ grep -o 'ERROR' app.log | wc -l
4
```
grep -o prints each match on its own output line, then wc -l counts those match lines. This pipeline is appropriate when the count should be occurrences instead of source lines.
Count matching lines without treating uppercase and lowercase text as different values.
```
$ grep -ci 'error' app.log
4
```
The -i option makes ERROR and error match the same pattern. The count is four because four lines contain that text after case is ignored.

Related: How to search case-insensitively with grep
Remove the sample file after testing.
```
$ rm app.log
```