Paralysis

Sleep

Stat Mods

The duration of sleep produced by an enemy's attack (as opposed to that produced by Rest) ranges from 1 to 8 rounds. The upper limit of this range depends on the relative Speed of the combatants, and on whether or not the "wake up" round is counted as a SLP round.

Sleep victims can wake up on their very next turn to attack. If the victim is slower than the "sleep-inducer", this means that it can wake up on the same round that it was put to sleep. It also means that the latest that it can wake up is on the 7th round, starting with the one on which the status change first occurred.

If the victim is faster than the sleep-inducer, it obviously can't wake up on the same round that it was put to sleep (because it was put to sleep at the end of the round). The latest it can wake up is on the 8th round (again, beginning with the one on which it was put to sleep).

Whatever the relative Speed, the maximum number of attack turns a Pokémon can lose due to SLP is 7, while the minimum is 1, and this is perhaps the best way to conceptualize sleep duration. It also provides a good reason to include the "wake up" round: one could split hairs about whether this actually counts as sleeping, but the practical effect is the same - your Pokémon can't attack.

Within this range, sleep duration appears to be randomly determined (1).

The duration of sleep produced by an enemy's attack ranges from 1 to 4 rounds. As in the RBY games, the exact duration of sleep depends upon the relative Speeds of the combatants, and on exactly which rounds are counted as SLP rounds. In practical terms however, the maximum number of attack turns a Pokémon can lose due to SLP is 3.

Sleep attacks will always fail if the victim currently has a Substitute.

Only Exeggcute learns more than one sleep attack. It starts out with Hypnosis, and learns Sleep Powder at L48. Though the latter move has 5 less PP, it is considerably more accurate, and should replace Hypnosis when the opportunity arises.

Since no other Pokémon learn more than one sleep attack (and there is no TM for any of them), the only decision you'll need to make regarding these moves is whether they're a worthwhile part of your Pokémon's moveset. In all cases, remember that Pokémon can only be affected by one major status change (SLP, PAR, PSN, FRZ, BRN) at a time (this does not include minor status changes like confusion, however). You may therefore wish to think carefully about letting your Pokémon learn a sleep attack if you also want it to know Toxic or Thunder Wave, for example.

Spore is the most accurate sleep attack, and is far and away the best reason to train a Parasect. Lovely Kiss and Sleep Powder are tied for 2nd most accurate, and should be considered very seriously for the Pokémon that learn them.

Hypnosis suffers a sharp drop in accuracy relative to the attacks mentioned so far, but even so it is probably a useful addition to most movesets. Gengar, in particular, can wreak havoc with Hypnosis on account of his superb Speed, and both he and Hypno can complement this move with Dream Eater: the most powerful Psychic attack and a useful healing move. Poliwrath has neither of these advantages, but is well suited to Hypnosis for a far more important reason: that cool swirl on his chest!

Sing is the least accurate of all sleep attacks. What's more, all of the Pokémon that learn this move also learn an excellent array of other attacks, either naturally or via TMs. It's hard to imagine a Jigglypuff worth the name that did not sing its enemies to sleep, but in general you should weigh the value of Sing carefully against more reliable attacks.

None.

To test this statement, a L50 Ditto and a L100 Mewtwo were put to sleep 140 times each. For the Ditto, the distribution of wake-ups across all 7 rounds was relatively equal. The Mewtwo showed a slightly greater frequency of wake-ups in the 1st and 2nd rounds of SLP than for the other rounds. However, a statistical test of significance indicated that the difference in frequency was well within the range of what should be expected from a random distribution. For that matter, a greater than average frequency of wake-ups was also found for the 5th round of SLP, but presumably this was not a function of Mewtwo's relatively high stats.

In all 280 of the trials described above, the subjects were put to sleep by a L50 Ditto. To investigate the possibility that sleep duration may be related to characteristics of the sleep-inducer, another 60 trials were done (against the Mewtwo) using a L3 Gastly. The distribution was very close to those obtained in the previous tests.

Even if there is an interaction between sleep duration and one or more characteristics of the victim, it would seem to be a very slight one. This page will be updated as more data become available.