Playing with a Full Deck

@Rick
Interesting link. It gives a better insight in the working of the macro.

@Willy,

Ah yes, I had forgotten that. The original routine that I developed the code I posted from was an array randomizer where the values were assumed to already be loaded in the array. Those values could be anything (random or ordered numbers, text or a mixture of both) and not simply numbers in numerical order as this article assumed. My original routine is quite old now... I originally published it in an old newsgroup (the predecessors to modern forums) devoted to VB (the compiled version, not the DotNET version) questions. I know the year because my code was incorporated into a web article back then which has a posting date on it of October 18, 1999...

http://vbnet.mvps.org/code/helpers/randomarray.htm

Randy, the man who maintained the website, used my RandomizeArray subroutine as part of a generalized presentation code of his own.

@Rick
Allen's macro as well as yours and my variant don't scramble the cells, they simply fill the selection with a random list overwriting it's contents. So you you don't have to fill the list with numbers first, you can as well start with an empty selection.
With Nums=Selection and a two dimensional array you get something new: shuffeling the selection maintaining their contents, whatever it is. This can be very usefull too.

@Willy,

Why would you want to randomly scramble empty cells? The idea behind my original code (which your modification retains) is to randomly rearrange the values within a selected columnar range of cells filled with values. For example, put the numbers 1 through 52 in cells A1:A52, select the range and run the code. if you think of the range as a deck of 52 cards, run the code and it shuffles them, run the code again and it shuffles them again. Of course, the code can randomly reorder any type of values within a columnar range of any size.

@Rick
Nums=Selection doesn't work. The selection being empty, It produces also an empty array, Or am I missing something?

@Willy,

Good idea to use a two-dimensional array instead of my one-dimensional array in order to remove the limitation built into the TRANSPOSE function when called from the Evaluate function. One note on that though, instead of loading up the Num array this way...

Nums = Evaluate("ROW(1:" & Selection.Count & ")")

you can simply do this (since Num is declared as a Variant) and end up with the identical array...

Num = Selection

Not sure if that will make the macro faster or not, but it does simplify the creation of the array.

Rick Rothstein's solution is a gem as usual. He claims his macro will be much faster than Allen's but didn't test it. I did and it is about 50 times faster !!! But while testing, I discovered his macro has a minor limitation: it has a limit of 65537 cells :-)
Further investigation revealed that TRANSPOSE is responsible for that. So I tried to produce the Num array only with ROW and it worked though the array is now two dimensional. I further took UBound(Nums) out of the loop because it doesn't change during the execution of the macro anyway. This makes the code in the loop a little bit simpler. Here is the result:

Sub FillRand()
Dim X As Long, RandIndex As Long, Nums As Variant, Cell As Range
Randomize
Nums = Evaluate("ROW(1:" & Selection.Count & ")")
X = UBound(Nums)
For Each Cell In Selection
RandIndex = Int(X * Rnd + 1)
Cell.Value = Nums(RandIndex, 1)
Nums(RandIndex, 1) = Nums(X, 1)
X = X - 1
Next Cell
End Sub

I did a test in Excel 2007 for a whole column and it took only 16 seconds to process the more than a million cells.

A simpler method without using macros is to populate one column with the number 1 thru 52 then in an adjacent column use the RND function to put a random number against each row.

Sort both the columns using the 'random' column as the key and you've randomised the order.

As in the example above you could end up with the same random number in the sort key but that does not matter.

I should have mentioned in both of my previous messages that the code I posted works in XL2003 (the earliest version I own) to XL2010 (the lastest version I own)... I see no reason why the code would not work in XL2013 or later, but I have no idea if it will work in versions prior to XL2003.

I constructed the code in my previous message just before going to sleep... when I awoke just now, it occurred to me that I could reduce the code to a single loop... excluding the Dim statement, my modified macro (below) is now down to eight active lines of code.

Sub FillRand()
Dim X As Long, RandIndex As Long, Nums As Variant, Cell As Range
Randomize
Nums = Evaluate("TRANSPOSE(ROW(1:" & Selection.Count & "))")
For Each Cell In Selection
RandIndex = Int((UBound(Nums) - X) * Rnd + 1)
Cell.Value = Nums(RandIndex)
Nums(RandIndex) = Nums(UBound(Nums) - X)
X = X + 1
Next
End Sub

Here is another (more concise) way to write the macro...

Sub FillRand()
Dim X As Long, RandIndex As Long
Dim Temp As Variant, Nums As Variant
Dim Cell As Range
Randomize
Nums = Evaluate("TRANSPOSE(ROW(1:" & Selection.Count & "))")
For X = UBound(Nums) To 1 Step -1
RandIndex = Int(X * Rnd + 1)
Temp = Nums(RandIndex)
Nums(RandIndex) = Nums(X)
Nums(X) = Temp
Next
For Each Cell In Selection
X = X + 1
Cell.Value = Nums(X)
Next
End Sub

Also note that the selection does not have to be contiguous so long as the parts that make it up do not overlap in any way.