But what is risk, exactly? Let's start by suggesting that job one of a game is to make people care. Job two is threatening all they care about, and herein lies the risk.
What can we possibly care about more than life itself. barring family, obviously? And in a tabletop game, our imaginary "life" counts for a lot. Heroes stake life and limb in a vast underworld and have a vested interest in staying alive. You know, risk. But what does death mean in a game where you just roll up another character and rejoin the adventure almost immediately? Especially when they inherit money and/or powerful magic items.
I once played no less than six dwarven brothers who picked up where their dead relatives left off, inheriting most of their stuff in the process. It was a brazen exploitation of the rulebook and something close to "immortality" in a game everyone calls lethal. No, what players really care about is the preservation of their cherished in-game personality figure...
And there's lots of ways to threaten these. Lots. Que maniacal laughter. Luvar the Shining gets turned into a slimy toad. Thingol the Foe Smasher loses his Dwarven Hammer. The diabolical possibilities are endless, and players will vigorously avoid them all. Herein lies yet another way to threaten the characters. Don't kill 'em. Let 'em live and force the players to struggle with an uncomfortable blow to the identity they've struggled to preserve.
With this in mind, is it possible to reduce in-game death while simultaneously raising the stakes and building greater dramatic tension? Ironically, letting characters live longer could actually make death more effective when it finally happens because the players have had time to grow attached. Talk about cold. Could it be that old-school dungeons might actually go easier on the players by killing off their heroes before attachments form?
That being said, here's an outline for an alternate deathless D&D where characters live longer but possibly suffer more. The latter is debatable; but the players will feel like all they hold dear is threatened, which feels a lot like death. Take it as the suggestion it is...
(1) When a character falls in battle, they go unconscious instead. Should at least one companion survive or manage to drag their friend to safety, that character revives and even recovers up to 10 lost hit points. It takes a total party kill to produce true death.
(2) Death magic, poison, and/or traps are still fatal, as is death from one-on-one fights with no backup. There's safety in numbers, and splitting the party or going alone is a bad idea.
(3) Cheating death incurs an experience debt equal to the opponent's level x 1,000, with earned experience applied to the deficit first. The player must pay off the debt before further advancement is possible, and since levels are everything, this one hits home!
(4) If advancement is capped, the character loses 1d3 levels instead. Alternately, they could sacrifice one or more magic items with a total experience value equal to the experience deficit incurred under rule #1, above. Higher-level characters have more options and more opportunities to recover, so the cost is firm (but fair) given that life itself is at stake.
Once again, these are just suggestions, and admittedly not fully formed ones!
Death is a clear and present danger, which in turn motivates careful planning to avoid the possibility. At the same time, too much death lessens the impact, which in turn saps its motivational power. It could be that improving survival while dishing out harsher, albeit non-lethal, consequences for defeat might actually feel riskier. Ultimately, this depends on the system and the desired ends of the group; but when story matters, so does survival.