There are a number of accounts within the KMyMoney data that are incomplete. Meaning, when the imports were done several years ago, not all of the accounts were imported. Most of those (incomplete) accounts had transactions from other accounts of course (double entry). Just doing a test on one of these accounts. Have exported from MS Money on Win8.1 (looks like 'loose' format is the one to use) and exported the same account from KMyMoney, then comparing the data with Beyond Compare.

Comparing the exports. The first set is from MS Money data, the second one from KMyMoney. Only the date format is different, that can be set on a QIF profile from memory.

!Type:Bank
D3/11'2004
T0.00
CX
POpening Balance
L[ANCU - Mickey Mouse]
^
---------------------------
!Type:Bank
D11.03.2004
T0.00
CX
POpening Balance
L[ANCU - Mickey Mouse]
^

The next set of data. The first set is from MS Money data, the second one from KMyMoney.

D3/11'2004
CX
MCash & Big Brother
T220.00
PCash/cheques
L[Petty Cash]
S[Petty Cash]
ETransfer - to open account
$160.00
SOther Income:Gifts Received
EMr & Mrs Fred Basset
$60.00
^
---------------------
^
D11.03.2004
MTransfer - to open account
T160.00
PCash/cheques
L[Petty Cash]

MS Money appears to add 'CX' transactions. Will KMyMoney ignore these ? (MS Money also has an 'N2' and other 'N.' transactions.)
The 'P' and 'L' transactions match.
The total 'T' transactions don't match but that is simply because the data in KMyMoney isn't complete.
What will the KMyMoney import do with the extra transactions, namely the 'S' and 'E'
If some transactions affect the (for example) Petty Cash account, will the import pause and allow an option ?
What happens with any duplicate transactions ?
What transaction type/s from MS Money contain the account identifier ?

Now of course the last question could be the most important one ..... Do I need to check/reconcile (possibly delete some QIF data from MS Money) prior to the KMyMoney import, or will KMyMoney magically handle it all and/or give me options at import ?

Can someone reply to these questions please ?

Also, when I previously went to 'reconcile' the QIF trans years ago with KMM, it was very helpful that the KMM transactions were in the same order as the QIF exported from MS Money. However, now they seem random, which isn't helpful at all.

Can I simply use an older version of KMM just for this exercise ?

Ooops, I am sorry, but I have never seen your original post. Regarding the randomness, I answered on this post.

QIF is by no means standardized. What I could expect without looking at the source of the QIF importer (which in some parts is older than 20 years), I could imagine that the duplicate usage of the account name Petty Cash in

L[Petty Cash]
S[Petty Cash]

is causing the problem you encounter (not sure though). The L-Record usually denotes a normal transaction referencing two account: the one you import and the one mentioned in the L-Record. The S-Record on the other hand is used in split transactions to identify an account. To add to the confusion, the brackets identify that it is an account. If they are not present, the name denotes a category.

BTW: CX simply is a C-Record with the setting X which means that this transaction is cleared and KMyMoney should not ignore them.

Thanks @ipwizard for your reply, much appreciated. And for the explanations. It seems the only way I'm going to reconcile the QIF exports is to:

1. Manually 'line up' a QIF export that is current with the QIF export from 2007 data. (with Beyond Compare)
2. Do some manual editing. (not too hard as the QIF data is in the same sequence)
3. This can only be done on an account by account basis.
4. After the appropriate backups to do the QIF imports.
5 Do a QIF export on the updated account and compare with 2007 data.

Fortunately, this is one of those work in progress jobs that has no urgency.

As the XML data issue was solved ( see viewtopic.php?f=69&t=175883 ), ran a QIF import today, using the old data from 2007. Was able to reconcile the QIF files, and do a manual check on the KMM data, before and after snapshots in XML format made it easier (with the environment variable QT_HASH_SEED set to zero.