On the vagueness of effect spells vs. area spells
Like many barrier or wall spells, the 6th-level cleric spell blade barrier [evoc] (Player's Handbook 205) seems to assume A) that creatures uniformly possess a size category of Medium and B) that a player will have his PC cast the wall spell so that it aligns with the map's grid lines. These are, of course, terrible assumptions to make in game with A) the word dragons in its title and B) actual players.
Anyway. According to the spell description, upon the blade barrier spell effect coming into existence in a square or grid intersection occupied by a creature, the creature "takes damage as if passing through the wall[, but it] can avoid the wall (ending up on the side of its choice) and thus take no damage by making a successful Reflex save." And, after that blade barrier spell effect has come into existence, a "creature passing through [it] takes 1d6 points of damage per caster level (maximum 15d6), with a Reflex save for half damage."
Now, I know it's risky, but I think it's safe to say that generally the way things should happen is that a creature that fails its saving throw against a newly created blade barrier effect and opts to remain in its current squares (even though they are now filled with whirling blades of force) should, later, be dealt damage (getting a Reflex saving throw for half damage). Then the question becomes When should the creature be dealt damage? and there's no good answer for that.
My understanding is that the closest thing there to an explanation of what may actually happen is in the Player's Handbook on Duration on Area that, in part, says
If the spell creates an effect, the effect lasts for the duration [i.e. 1 min./level for the blade barrier spell].… If the spell affects an area, as silence does, then the spell stays with that area for its duration. Creatures become subject to the spell when they enter the area and are no longer subject to it when they leave. (176 and emphasis mine)
Thus, using a what I believe is the typical reading of the word enter, according to one possible application of this rule—and this application is, to be clear, imprecise (see below)—the creature would not be dealt damage again if it stays in the same squares that it was in when the blade barrier spell effect arrived. Further, the creature isn't dealt damage again if it continuously keeps at least one of its squares within the blade barrier spell effect. Technically, the creature did not enter the spell effect but remained in the spell effect, has already been subject to the spell effect, and can't be subject to it again until it exits and reenters.
Kram casts the blade barrier spell so that the spell's effect occupies some of Yor's space. Yor makes a Reflex saving throw against the effect. Failure means both that Yor is dealt the effect's damage and that he remains where he is. Success means both that Yor is not dealt the effect's damage and that Yor must move so that none of the squares or grid intersections he occupies is also occupied by the blade barrier spell effect. Yor fails the saving throw, is dealt damage, and remains in his squares.
On Yor's turn, as he hasn't exited all of the squares that contain a blade barrier effect, Yor is unaffected by the spell's effect. Yor can exit or move within the area that Yor shares with the spell effect freely, but if Yor at any point finds that his space no longer occupies at least one square that's also affected by the blade barrier spell effect, when he reenters a square that's affected by the blade barrier spell effect he'll makes a Reflex saving throw against the blade barrier spell effect, success meaning that he's dealt half damage and failure meaning that he's dealt full damage.
Obviously, narrative justifications for a blade barrier spell effect not dealing damage to the typical creature that opts to stay in it tend to ring hollow, but this is how things go in campaigns when I'm the DM. This is also how other DMs I've had ruled on this issue. So far as I'm aware, there's just no rules support for a different reading. (I honestly look forward to other answers in hopes that someone's discovered something I've overlooked!) Other groups are likely to handle this differently, but groups I'm in resist changing the rules unless the rules are causing actual problems… and despite these rules being weird and lacking verisimilitude, they are playable. So we carry on.
If uninterested in game engine diagnostics, you're done here. Good gaming!
Why the game's assumptions are bad assumptions
First, all creatures don't possess a size category of Medium. If they did, then getting from one side to the other of a vertical two-dimensional effect is pretty easy. However, if a corner square of a creature that possesses the size category Colossal is caught in the effect, a successful saving throw can see the creature instantaneously and off-turn move 50 ft. closer to the caster! This consequence is implied in Skip Williams's Rules of the Game Web column "All about Movement, Part Seven" on Barrier Spells that, in part, says
Some barrier spells, such as wall of fire and blade barrier deal damage to things that pass through them, but they do not otherwise block movement unless their descriptions say they do. Most such spells can be aimed so that they appear in the same location as a creature. If so, the creature takes damage as though it passed through the barrier. Usually a creature that has a barrier dropped right on it in this manner can attempt a Reflex save to avoid damage. If this save succeeds, the creature avoids the barrier and ends up on the side of its choice (this occurs at the time the barrier occurs and doesn't count against the creature's movement during its next turn).
(Emphasis mine.) This may have players crying foul, especially since bigger creatures already have enough advantages. A successful saving throw should require the creature to exit all of the squares it occupies that are also occupied by the effect by the shortest route possible rather than a route the creature picks. (However, it seems this movement provokes attacks of opportunity normally, so there's that.)
Second, what makes the Player's Handbook rule above imprecise is that it's describing spells with an area entry and not spells with an effect entry. The point of origin of a spell with an area entry is always a grid intersection, but the caster of a spell with an Effect entry (like the blade barrier spell's entry Effect: Wall of whirling blades up to 20 ft. long/ level…) can put that spell's effect anywhere to which the caster has line of sight and line of effect and that the spell allows… and in the case of wall spells, that means they need not be placed so that they align neatly along grid intersections. That some wall spells are also two-dimensional affairs only adds to this problem. "Reading Spell Descriptions, Part Five" (another Rules of the Game Web column) on Wall Spells, in part, confirms this problem, saying that
Spells such as wall of fire are effect spells that create barriers. Often, the effect entry for such a spell does not mention how thick the barrier is (for example, blade barrier, wall of fire, and wall of force). In these cases, the wall is basically two-dimensional, with only a negligible thickness. If such a spell deals damage (or has some other effect), the spell deals damage to anything that passes through it or otherwise breaks the plane of the barrier. In the case of a wall of force, nothing can break the plane (save for a few magical effects specified in the spell description).
A wall that's a line (a game line—a series of 5-ft. squares in row) is relatively easy to manage on the grid, but a wall that's a flat plane is much harder to practically implement. When a game element says that something occupies less than a square, determining how many of of them fit in a square is easy, but when a game element splits squares—potentially at crazy angles—, determining how many of, how much of, or even if something fits in that divided square is an argument waiting to happen.
(Also, while there might seem to be some saving grace in the idea that If such a spell deals damage[, it] deals damage to anything that passes through it or otherwise breaks the plane of the barrier, but a creature that sees the barrier evoked in its space then remains stationary or moves within the barrier has already broken the plane of the barrier. That creature can still remain there safely!)
Note: I tried to imagine a narrative justification for being able to hang out unscathed in a blade barrier spell effect, and the best I could do is say that blade barrier spell effects and similar spell effects have a pattern that can be discerned only by experiencing the spell's effect; exiting and reentering the effect requires discerning the pattern anew. That's eccentric but not awful, I guess.