Just an addition to my previous note: In unicode mode (with "u" modifier), PREG_BAD_UTF8_ERROR only reflects errors in the subject string. If the pattern itself contains invalid characters, preg_match() (or preg_match_all()) returns false but preg_last_error() returns 0 indicating PREG_NO_ERROR. Instead, php issues a warning: "preg_match(): Compilation failed: invalid UTF-8 string at offset 0"