Arabic verbs (فِعْل fiʿl; pl. أَفْعَال afʿāl), like the verbs in other Semitic languages, and the entire vocabulary in those languages, are based on a set of two to five (but usually three) consonants called a root (triliteral or quadriliteral according to the number of consonants). The root communicates the basic meaning of the verb, e.g. ك-ت-ب k-t-b 'write', ق-ر-ء q-r-ʾ 'read', ء-ك-ل ʾ-k-l 'eat'. Changes to the vowels in between the consonants, along with prefixes or suffixes, specify grammatical functions such as person, gender, number, tense, mood, and voice.

Various categories are marked on verbs:

Weakness is an inherent property of a given verb determined by the particular consonants of the verb root (corresponding to a verb conjugation in Classical Latin and other European languages), with five main types of weakness and two or three subtypes of each type.

Arabic grammarians typically use the root ف-ع-ل f-ʿ-l to indicate the particular shape of any given element of a verbal paradigm. As an example, the form يتكاتب (root: ك-ت-ب) yutakātabu 'he is corresponded (with)' would be listed generically as يتفاعل yutafāʿalu (yuta1ā2a3u), specifying the generic shape of a strong Form VI passive verb, third-person masculine singular present indicative.

The maximum possible total number of verb forms derivable from a root — not counting participles and verbal nouns — is approximately 13 person/number/gender forms; times 9 tense/mood combinations, counting the س- sa- future (since the moods are active only in the present tense, and the imperative has only 5 of the 13 paradigmatic forms); times 17 form/voice combinations (since forms IX, XI–XV exist only for a small number of stative roots, and form VII cannot normally form a passive), for a total of 1,989. Each of these has its own stem form, and each of these stem forms itself comes in numerous varieties, according to the weakness (or lack thereof) of the underlying root.

Each particular lexical verb is specified by four stems, two each for the active and passive voices. In a particular voice, one stem (the perfective stem) is usually used for the past tense, and the other (the imperfective stem) is usually used for the present and future tenses, along with non-indicative moods, e.g. subjunctive and imperative. Though there is still some disagreement about the interpretation of the stems as tense or aspect, the dominant current view is that the stems represent aspect, sometimes of a relative rather than absolute nature. In this system of classification, the ostensibly "past" and "non-past" stems are called the perfective stem and imperfective stem.

To the past stem, suffixes are added to mark the verb for person, number and gender, while to the non-past stem, a combination of prefixes and suffixes are added. (Very approximately, the prefixes specify the person and the suffixes indicate number and gender.) A total of 13 forms exist for each of the two stems, specifying person (first, second or third); number (singular, dual or plural); and gender (masculine or feminine).

There are six separate moods in the non-past: indicative, imperative, subjunctive, jussive, short energetic and long energetic. The moods are generally marked by suffixes. When no number suffix is present, the endings are -u for indicative, -a for subjunctive, no ending for imperative and jussive, ـَنْ -an for shorter energetic, ـَنَّ -anna for longer energetic. When number suffixes are present, the moods are either distinguished by different forms of the suffixes (e.g. ـُونَ -ūna for masculine plural indicative vs. ـُو -ū for masculine plural subjunctive/imperative/jussive), or not distinguished at all. The imperative exists only in the second person and is distinguished from the jussive by the lack of the normal second-person prefix ـت ta-/tu-.